Characterization of Cardiac Tissue Using Optical Coherence Tomography

CHARACTERIZATION OF CARDIAC TISSUE USING

OPTICAL COHERENCE TOMOGRAPHY

CHRISTINE P. FLEMING

Submitted in partial fulfillment of the requirements

For the degree of Doctor of Philosophy

Dissertation Adviser: Dr. Andrew M. Rollins

Department of Biomedical Engineering

CASE WESTERN RESERVE UNIVERSITY

May, 2010

CASE WESTERN RESERVE UNIVERSITY

SCHOOL OF GRADUATE STUDIES

We hereby approve the thesis/dissertation of Christine P. Fleming

candidate for the PhD degree *.

(signed) ______Andrew Rollins (chair of the committee)

______Igor Efimov

______David Rosenbaum

______Kenneth Singer

______Xin Yu

(date) ______

*We also certify that written approval has been obtained for any proprietary material contained therein.

Table of Contents Acknowledgements ...... 17 List of Abbreviations ...... 18 Abstract ...... 21 Chapter 1: Background and Significance ...... 24 1.1. Catheter Ablation for the Treatment of Cardiac Arrhythmia ...... 24 1.2. Radiofrequency Ablation (RFA) ...... 26 1.2.1. Monitoring and Guidance of RFA ...... 30 1.2.2. Animal Models for Ablation ...... 31 1.2.3. Fiber Orientation ...... 31 1.3. Optical Coherence Tomography ...... 33 1.3.1. Time Domain Optical Coherence Tomography ...... 36 1.3.2. Fourier Domain Optical Coherence Tomography ...... 38 1.3.3. Polarization Sensitive OCT ...... 43 1.3.4. OCT Catheter Probes ...... 44 1.3.5. OCT Imaging of the Cardiovascular System ...... 45 1.3.6. Image analysis ...... 47 1.4. Objective ...... 50 Chapter 2: Characterization of Arrhythmogenic Substrates Using Optical Coherence Tomography ...... 52 2...... 52 2.1. Introduction ...... 52 2.2. Two dimensional fiber orientation algorithm ...... 54 2.2.1. Results ...... 58 2.2.2. Healed Myocardial Infarction ...... 60 2.2.3. High Resolution Imaging ...... 63 2.3. Analysis of an Optical Mapping Inverse Model ...... 66 2.4. Discussion ...... 76 2.5. Conclusion ...... 78 Chapter 3: Characterization of ablation lesions using optical coherence tomography ...... 79

3...... 79 3.1. Introduction ...... 79 3.1.1. Radiofrequency Ablation ...... 79 3.1.2. Monitoring Radiofrequency Ablation Therapy ...... 80 3.2. Methods ...... 82 3.2.1. Feasibility Testing ...... 82 3.2.2. In vitro ablation protocol...... 89 3.2.3. Optical Coherence Tomography Imaging ...... 91 3.2.4. Validation ...... 92 3.2.5. Image Analysis ...... 92 3.3. Results ...... 96 3.3.1. Tissue Classification ...... 97 3.3.2. Visualization of Overtreatment ...... 104 3.4. Discussion ...... 106 3.5. Conclusion ...... 111 Chapter 4: Toward Guidance of Epicardial Radiofrequency Ablation Therapy using Optical Coherence Tomography ...... 112 4...... 112 4.1. Introduction ...... 112 4.1.1. Epicardial Ablation ...... 112 4.1.2. Pre-procedural imaging ...... 113 4.1.3. Optical Coherence Tomography ...... 114 4.2. Methods ...... 116 4.2.1. Sample Preparation ...... 116 4.2.2. Imaging Protocol ...... 117 4.2.3. Validation ...... 119 4.3. Results ...... 119 4.3.1. Discussion ...... 127 4.3.1.1. Substrate Characterization ...... 128 4.3.1.2. Real time Guidance ...... 129 4.3.1.3. Future Work ...... 130 4.3.2. Conclusion ...... 131

Chapter 5: Real-time Monitoring of Cardiac Radiofrequency Ablation Lesion Formation using an Optical Coherence Tomography Forward Imaging Catheter ...... 132 5...... 132 5.1. Introduction ...... 132 5.2. Methods ...... 134 5.2.1. Feasibility Testing ...... 134 5.2.4. Real Time Monitoring of Ablation Lesion Formation ...... 144 5.2.5. Imaging Protocol ...... 145 5.2.6. Validation ...... 145 5.2.7. Image Analysis ...... 146 5.2.8. Statistical Analysis ...... 148 5.3. Results ...... 148 5.3.1. Forward Imaging Probe ...... 148 5.3.2. Real time visualization of lesion formation ...... 150 5.3.3. Imaging in the presence of blood ...... 153 5.4. In vivo imaging using the forward imaging probe ...... 154 5.5. Discussion ...... 158 5.5.1. Forward Imaging Probe ...... 158 5.5.2. Real Time Visualization of Lesion Formation ...... 160 5.5.3. Assessment of Tissue Contact in the Presence of Blood ...... 160 5.5.4. In vivo imaging ...... 161 5.6. Conclusion ...... 163 Chapter 6: Summary and Future Work ...... 164 6...... 164 6.1. Summary ...... 164 6.2. Clinical Significance ...... 166 6.3. Future Work ...... 170 Bibliography ...... 178

Table 1. 1 Energy Sources for Catheter Ablation[6] ...... 25

Table 1. 2 Biophysics Equations for Radiofrequency Ablation ...... 26

Table 1. 3. Success and complication rates of radiofrequency ablation therapy by arrhythmia targeted[10]...... 29

Table 3. 1. Binary discrimination of lesions and untreated tissue ...... 102

Table 5. 1. Forward imaging OCT catheter design specification for optical characteristics, mechanical characteristics, and environmental conditions ...... 139

Table 5. 2. Spot size as a function of GRIN lens temperature ...... 149

Figure 1. 1. Ablation lesion formation using radiofrequency ablation energy. Lesions are formed by resistive heating with RF energy delivery. A rim, 1-2 mm, is heated directly by the catheter, zone of resistive heating. The rim acts as a heat source to heat deeper regions of the tissue, zone of conductive heating...... 27

Figure 1. 2. Michelson interferometer. A) Interferometer schematic. B) Interference pattern for light source with a narrow bandwidth. C) Interference pattern for light source with broad bandwidth. Interference occurs when the path length difference of the

sample, ls, and reference arm, lr, is within the coherence length, lc, of the light source. .. 34

Figure 1. 3 OCT image generation of mouse epicardium. (a) 1-D axial scan. (b) 2-D B- scan image generated by transverse scanning and collection of multiple axial scans. (c) 3- D volume reconstruction from raster scanning across surface, collecting a series of B- scan images...... 35

Figure 1. 4. Time Domain OCT system. Axial scans generated by mechanical scanning of reference mirror. Reflection sites in depth localized to within the coherence length of light source...... 37

Figure 1. 5. Fourier Domain OCT implementations. (a) Spectral Domain OCT (SDOCT). (b) Swept Source OCT (SSOCT) or Optical Frequency Domain Imaging (OFDI) ...... 39

Figure 1. 6. Axial scan generation with FDOCT. (a) Three reflection sites within sample. (b) Spectral interferogram represents summation of sinusoidal modulation of spectrum. (c) Axial scan obtained by computing the Fourier transform of a spectral interferogram that is evenly spaced in k...... 41

Figure 2. 1: 3D OCT data sets of the Right Ventricular Free Wall (RVFW). A) Three-dimensional OCT images of the RVFW, data sets A1) RVFW1, A2) RVFW2, A3) RVFW3. Arrow in A1 points to cross sectional (en-face) slices. B) View of three en-face slices in depth within 3D OCT data sets. Fiber structure is visible within en-face slices. The uneven sample surface (B1) and trabeculations (B2 and B3) induce low frequency changes and produce shadows within subsequent en-face slices, inducing a gradient that may be greater than the gradient produced by the fibers...... 55

Figure 2. 2. Pre-processing of en face images to reduce intensity gradients due to surface topology shadows and noise reduction. A) raw en face image. B) en face image after high pass filtering. C) en face image after high pass filtering and noise reduction with a Wiener filter...... 56

within the images due to uneven surfaces (A) and shadows created by trabeculations (B, C) the two-step filtering process effectively reduce their contribution to the gradient magnitude calculations...... 59

Figure 2. 4: Fiber orientation as a function of depth. A) Fiber orientation as a function of depth for RVFW1 comparing automated (filled square) and manual (open square) measurements for a 281 x 281 x 500µm volume. A1) Comparison of fiber orientation measurements for an 1125 x 1125 x 500 µm volume within RVFW1. Open square – RVFW1 manual, filled square – RVFW1 automated. Automated fiber orientation measurements agree very well to manual fiber orientation measurements. B) Fiber orientation as a function of depth for RVFW2 (filled diamond) and RVFW3 (filled triangle) for a 281 x 281 x 500µm volume. Fiber orientation changes monotonically with depth...... 60

Figure 2. 5: OCT volumes and fiber orientation within a healed myocardial infarction rabbit. A) Three dimensional reconstructions. B) Fiber Orientation within en face image. 1) Infarction, 2) Infarction / Healthy tissue, 3) Healthy tissue. Decrease in fiber organization within the infarction reflected by the broad fiber angle histogram ...... 62

Figure 2. 6: Representative OCT images from healed myocardial mouse infarction model from two hearts. Increased scattering and significance decrease in ventricular thickness observed within infarction. Adjacent viable tissue has birefringence artifact. a,d) healthy. b,e) border, transition zone. c,f) infarction. Scale box is 500µm x 500µm...... 63

Figure 2. 7: High resolution imaging of the mouse myocardium. A) Three dimensional reconstruction of wild type mouse right ventricle and right atria. B,C) Representative B-scans from volume. D) Slice parallel to epicardial surface, with overlay of fiber orientation vectors .... 65

Figure 2. 8 Flow diagram for optical mapping inverse model. 1) Computation of the illumination profile within the sample, Φillm, using a forward model. 2Computation of the membrane potential within the sample. 4. The flux of membrane potential at the surface is computed to approximate an optical signal that the camera records, Vopt(x,y). 5. The computed optical signal is compared to the optical mapping data. If the computed optical signal is not similar to the experimental recordings, kinetic parameters are updated, and steps 2-4 are repeated until a Vm(x,y,z) distribution produces a similar optical signal to the experimental data...... 67

Figure 2. 9. Nonlinear Least Square Algorithm for Reconstruction Depth Distribution of Membrane Potential. First step is the computation of Φillm(x,y,z) using finite difference approximation of the steady state diffusion equation for highly scattering medium using. Boundary conditions are derived from computational domain and the diffusion coefficient which is dependent on μs and g. The second step is the Computation of Vm(x,y,z) using the Beeler Reuter model. The diffusion equation is used to compute the emission signal, Φem(x,y,z), with the membrane potential and illumination as inputs. The calculated optical mapping signal, Vopt(x,y), is computed by taking the flux of the emission signal. The norm of the difference between the computed optical signal and the measured optical signal is taken to determine if

parameters, p, need to be updated which will modify Vm(x,y,z) that produces a similar optical signal to what is recorded with Optical Mapping ...... 68

Figure 2. 10. Condition number as a function of time for the matrix A (Levenberg Marquardt 3 method) ...... 73

Figure 2. 11. Condition number of matrix A3 as a function of regularization parameter (Levenberg Marquardt method) ...... 74

Figure 2. 12. Comparison of computed and initial parameters ...... 75

Figure 3. 1. Long working distance OCT scanner. A) Zemax optical design of scanner. B) Point spread function...... 83

Figure 3. 3: Classification of lesions and untreated tissue. Using signal decay rate and correlation coefficient, separation between ablation lesions and viable...... 87

Figure 3. 4: Analysis parameters as a function of lesion depth. Signal decay rate and correlation coefficient of ablation lesions statistically lower than viable tissue. Error bars are 95% confidence intervals...... 88

Figure 3. 5. Image analysis and standard functional measurements as a function of lesion depth. A) nearly linear decrease in light attenuation rate with increasing lesion depth. Rate of change of lesion depth plateaus for large lesions (>4mm) b) decrease in correlation coefficient with increasing depth. C) nearly linear increase in temperature with increasing lesion depth. D) slight decrease in impedance with increasing lesion depth. Lesion depth determined by TTC vital staining...... 88

Figure 3. 6. Visualization of adverse complications. OCT Image of endocardial lesion 1: 35W, 345Ω. Ablation created a steam pop on tissue surface after 3 seconds. Carters are observed within the OCT image...... 89

of ablative lesions) C) Optical coherence tomography (OCT) system. D) Representative OCT image of a seven millimeter image of ablative lesion and adjacent tissue E) TTC stains of ablative lesion shown in panel D)White necrotic tissue of ablative lesion and red viable tissue demonstrated ...... 91

Figure 3. 8. Dark band due to tissue birefringence a-b) OCT images of an untreated site obtained with different polarization states of the sample arm light. . Location of band moves as the polarization state in the sample arm is changed. Birefringence dependent bands are highlighted with green arrows. c) Representative OCT image of an ablation lesion. d-f) Decimated and flattened version of images shown in a-c. Region of interest, 525µm, used in analysis shown in panel e as white horizontal lines. g-j) Representative averaged axial scans from the sites indicated in e-h, shows change in location of the band within axial scans of untreated site (g and h). No band in OCT image of ablation lesion. Images acquired with time domain (TDOCT) system...... 93

Figure 3. 9: Image analysis to distinguish over treated lesions. A) Representative OCT image of an over treated lesion. B) Filled in version of image in (a) identifying voids in myocardium. C) Average intensity as a function of depth. Blue curve represents entire axial scan blue region from (a) starting at the sample surface. Red curve represents region of interest used to fit to linear line. Correlation of fit used to assess heterogeneity ...... 96

Figure 3. 10. A) Temperature duration curve. Catheter-tissue temperature increased monotonically over time, up to a maximal temperature of 70 degrees Celsius. B) Impedance measurements during RF ablation. Impedance measured at the catheter tissue interface decreased during RF applications, consistent with formation of lesions. C) Strength duration curve. Power was consistent throughout the RF applications at 25 Watts. D). Lesion development over time. The depth of the lesion created by RF increased with longer duration of RF application. Results +/- 95% CI ...... 97

Figure 3. 11. Gap of untreated tissue within linear ablation line characterized by band due to polarization artifact. a) The gap of untreated tissue was characterized by a strong birefringence dependent band. b) TTC vital staining of ablation two ablation lesions with gap of untreated tissue. c) OCT image decimated to 512 x 13 pixels. d-f) Representative averaged axial scans from decimated image within areas with lesion (d,f) and untreated tissue (e). Band is observed within area of treated tissue and not ablation lesions. Images acquired with microscope integrated Fourier Domain (FDOCT) system...... 98

line indicates location of B-scan shown in panel B. B) OCT B-scan demonstrates absence of birefringence band in necrotic tissue of RF lesion. Birefringence band is present on the left, within the area of untreated tissue; whereas, dark band is absent within area of lesion. Birefringence band is also demonstrated in double hump within average axial scan (red) in the area with no lesion. C) Slice parallel to the sample 480µm below sample surface, showing fiber organization within area without a lesion and lack of fiber organization within the lesion. Images acquired with microscope integrated Fourier Domain (FDOCT) system...... 99

Figure 3. 13. Representative OCT images of untreated ventricular endocardium. Birefringence band is visible within images of untreated tissue (indicated by green arrows in panel A). Right ventricle (A,B), left ventricle (C,D), and right ventricular septum (E,F). Images acquired with time domain (TDOCT) system...... 100

Figure 3. 14. Representative OCT images of endocardial radiofrequency ablation lesions. Pins are placed along the ends of the lesion and are visible within OCT images. Yellow dotted lines indicate the area of the RFA lesion. Ablation lesions are characterized by increased signal intensity and absence of polarization artifact. Right ventricle (A,B), left ventricle (C,D), and right ventricular septum (E,F). Images acquired with time domain (TDOCT) system...... 101

Figure 3. 15. Distinguishing ablation lesions and untreated tissue using gradient strength. a) Scatter plot for each image within the TDOCT dataset, with gradient strength on the x- axis and intensity on the y-axis. Ablation lesions (open shapes), untreated tissue (filled shapes). Left ventricle – circle, right ventricle – square, ventricular septum - triangle B) Receiver operator characteristic (ROC) curve for gradient strength to distinguish ablation lesions from untreated tissue. Lesions can be distinguished from untreated samples within all sites using gradient strength as a discriminating factor, with a 0.94 area under curve (AUC). C) ROC curve for intensity. Intensity had a low classification power to distinguish ablation lesions from untreated tissue, with a 0.72 AUC...... 103

Figure 3. 16. Representative OCT images of the endocardium with visualization of “over treated” RFA lesions. Disruptions within the endocardium and myocardium are visible, and may be precursors to steam pops and crater formation. Yellow dotted lines circumscribe each lesion. Images acquired with time domain (TDOCT) system...... 104

Figure 3. 17. Over treated lesions distinguished from adequate lesions by correlation and morphological filling. Adequate lesions are clustered on the bottom corner of plot. Overtreated lesions characterized by increased heterogeneity (increase correlation coefficient) and voids in the myocardium (increased intensity difference after filling). 105

Figure 3. 18. Representative OCT images of left atrial radiofrequency ablation lesions. a- c) Representative images of untreated left atrial Endocardium. Left atrial contains complex structure with areas that are pectinated (a) and smooth (b, c). The endocardial layer varies in thickness. Birefringence band is not present within untreated atrial images. d-f) representative OCT images of RFA lesions within left atrial appendage. Ablation lesions shown decrease in contrast between endocaridum and myocardium. . 106

Figure 4. 1. Experimental Samples. Gross view of epicardial surface of a wedge of (a) right ventricle and (b) left ventricle. Yellow arrows indicate sub-epicardial fat, black arrows indicate coronary vessels, white arrows indicate radiofrequency ablation lesions...... 117

Figure 4. 2. Optical schematic of forward imaging catheter probe. GRIN lens used for focusing, Risley prism deflects beam off axis, optical glass isolates optics from environment. Beam is focused 0.5mm from the optical glass. Fiber glued to GRIN lens– Risley prism unit. Application of torque on fiber creates cone scanning...... 119

Figure 4. 3. Representative images of normal myocardium imaged from epicardial surface of the left ventricle. (a, b) microscope images of epicardial surface correlating to acquired three dimensional OCT image sets. Black square indicates field of view, 4mm by 4 mm. (c) Example B-scan image from volume represented in (a). (d) Example B- scan image from volume represented in (b). Epicardium appears as a thin bright layer (green arrow). Dark band (red arrow) within myocardium results from birefringence properties of the myocardium. Scale box is 500μm by 500μm...... 121

Figure 4. 4. Representative images of epicardial fat imaged from the right ventricle of two hearts. (a, b) microscope images of epicardial surface correlating to acquired three dimensional OCT image sets. Black square indicates field of view, 4mm by 4 mm. (c) Example B-scan image from volume represented in (a). (d) Example B-scan image from volume represented in (b). Beneath the layer of epicardial cells and supporting connective tissue (green arrow) is the epicardial fat (yellow arrow), consisting of adipose tissue that appears heterogeneous within OCT images. Scale box is 500μm by 500μm. 122

Figure 4. 5. Representative images of coronary vessels from the left and right ventricle. (a, b) microscope images of epicardial surface correlating to acquired three dimensional OCT image sets. Black square indicates field of view, 4mm by 4 mm. (c) Example B- scan image from volume represented in (a). (d) Example B-scan image of volume represented in (b). Coronary vessels appear as signal poor regions (black arrow), corresponding to the vessel lumens. Scale box is 500μm by 500μm...... 123

Figure 4. 6. Representative images in heterogeneous tissue regions, including myocardium (red arrow), epicardial fat (yellow arrow), and coronary vessels (black arrow). (a) Microscope image of coronary surrounded by epicardial fat. (b) OCT slice parallel to sample surface, 230μm below surface, corresponds well to microscope image (a). (c) Microscope image of coronary in fat and myocardium. (b) OCT slice parallel to sample surface, 200 μm below surface, corresponds well to microscope image (a). (e, g) Example B-scan images from volume represented in (a, b). Thick epicardial layer covers epicardial fat, which surrounds coronary vessel. (f, h) Example B-scan images from volume represented in (c, d). Volume acquired for area encompassing epicardial fat, coronary, and untreated myocardium. (f) B-scan encompasses untreated myocardium characterized by polarization artifact band (red arrow) and epicardium (green arrow) and coronary (black arrow). (h) B-scan encompasses epicardial fat (yellow arrow), which appears heterogeneous, epicardium (green arrow) and coronary (black arrow). Scale box is 500μm by 500μm...... 124

Figure 4. 7: Representative images of epicardial radiofrequency ablation lesion within the right ventricle. (a) Microscopic image of field of view in which volumetric OCT image was obtained, black box 4mm by 4mm. (b) TTC vital staining confirming necrosis, where a transmural lesion was generated. (c, d, e) example OCT B-scans encompassing the ablation lesion (white arrow) and adjacent untreated tissue (red arrow). Untreated tissue characterized by contrast between epicardial layer (green arrow) and myocardium and polarization dark band artifact. Ablation lesion, necrotic tissue characterized by absence of contrast between epicardial layer and myocardium and absence of polarization dark band artifact. Scale box is 500μm by 500μm...... 125

Figure 4. 8. Representative images of epicardial substrates with the forward imaging catheter probe. (a) untreated myocardium in contact. (b) epicardial fat in contact (c) epicardial fat not in contact. (d) coronary vessel. Red arrow pointing to birefringence band within untreated myocardium, green arrow pointing to epicardial layer, yellow arrow pointing to epicardial fat, black arrow pointing to coronary vessel. Scale bar 500μm...... 127

Figure 5. 1. Experimental setup for imaging real time dynamics of ablation lesion formation using a bench top OCT scanner...... 134

Figure 5. 2. Real-time imaging of radiofrequency ablation. RF probe located at far right of image. Arrows point to ablation lesion. Within 2 seconds, ablation lesion is being formed. B-scan image 8mm. Ablation parameters: 35W, impedance 105Ω, time 13seconds. Images acquired with a TDOCT system...... 135

Figure 5. 3. Coagulum formation during ablation. Ablation parameters: 20W, impedance 312Ω, time 7seconds. Yellow arrow pointing to lesion and green arrow pointing to coagulum. Craters created by ablation. Craters are observed in location of center of ablation catheter. A) OCT image before RF energy. B) Post ablation image, after 7 seconds at 20W. Sample shifted by 3mm to show crater created by RF energy. Yellow arrow pointed to lesion, green arrow pointed to crater. B-scan 8mm, averaged 9 frames. Images acquired with a TDOCT system...... 136

Figure 5. 4. Forward imaging optical designs. A) axial imaging. B) Circular imaging. C, D) spot diagrams for the axial and circular designs respectively. E,F) Huygens point spread function (PSF) for the axial and circular designs respectively. G) Cross sections of point spread function in the x and y direction showing a 30µm FWHM for both the axial and circular probes. Center of PSF shifts for the circular probe, indicating presence of aberrations...... 138

Figure 5. 6. System design and specification. A) SDOCT. B1) Bench-top sample arm B2) Forward imaging probe sample arm. C) Spectrum of light source. D) SNR as a function of depth. E) Point spread function to measure axial resolution...... 143

Figure 5. 7: Imaging Quality of Forward Imaging Catheter Prototypes. Non-uniform rotation rate within the catheter probe prototypes are reflected in streaked appearances (yellow arrows) within OCT images. Severity of this problem varied from probe to probe. Probe CW08 (c) and CW10 (d) have significant portions of the image with streaked appearances. Probes CW06 (a) and CW07 (b) were used for further experiments...... 144

Figure 5. 8. Experimental Protocol. A) Samples placed in custom chamber with superperfusion flow of PBS. Real time dynamics recorded with forward imaging catheter bound side by side to RFA catheter. b) OCT imaging conducted over 90 sections and 60 seconds of RF energy delivery...... 145

Figure 5. 9. Image visualization of catheter images. a) Forward imaging probe uses GRIN lens for focusing and Risley prism for beam deflection. b) OCT image obtained by

forward imaging probe. c) Zoomed in region from OCT image in (b) showing areas with streaking. d) OCT image after correlation based algorithm for correcting non-uniform scanning. e) Zoomed in region from OCT image in (d) showing removal of streaking artifact. f) Decimated version of corrected OCT image from (d) to be used for image analysis. Example averaged axial scan is plotted in red. g) Visualization of OCT image from (d) displayed with correct aspect ratio. Image acquired with CW07 and FDOCT system...... 147

Figure 5. 10. Representative images from forward imaging probe. a) In vivo images human skin. Layers and sweat ducts are visible. b) Formalin fixed swine right ventricular epicardium with visible coronaries. Scale bar 500µm...... 149

Figure 5. 11. Visualization of dynamics due to RF energy delivery. a, c, e, g) representative OCT images at 10 seconds, 20 seconds, 70 seconds, and 90 seconds during experimental protocol. Baseline images characterized by birefringence band. Over time, birefringence band disappears and intensity increases. b, d, g, h) representative gradient strength images for corresponding intensity images in a, c, e, and g. Gradient strength computed by filtering intensity image with a LoG kernel. Birefringence band appears white within gradient strength images. i) image intensity increases nearly linearly with RF energy delivery. j) gradient strength decreases nearly linearly with RF energy delivery. k) TTC vital staining stains necrotic tissue white and viable tissue red. Scale bar 500µm ...... 151

Figure 5. 12. Visualization of dynamics due to RF energy delivery. a, c, e, g) representative OCT images at 10 seconds, 20 seconds, 70 seconds, and 90 seconds during experimental protocol. Baseline images characterized by birefringence band. Within the application of RF energy, location of birefringence bands lower and then disappear. . b, d, g, h) representative gradient strength images for corresponding intensity images in a, c, e, and g. Gradient strength computed by filtering intensity image with a LoG kernel. Birefringence band appears white within gradient strength images. i) image intensity increases nearly linearly with RF energy delivery. j) gradient strength decreases nearly linearly with RF energy delivery. k) TTC vital staining stains necrotic tissue white and viable tissue red. Scale bar 500µm ...... 152

Figure 5. 13. Assessment of tissue contact in the presence of blood. Samples submerged in heparized blood. a) Image obtained when probe was in direct contact with endocardial surface. b)Imaging depth significantly decreased when probe was not in direct contact with endocardial surface. Attenuation primarily due to blood absorption and scattering. Scale bar 500µm ...... 154

Figure 5. 14. In vivo cardiac imaging in a swine. a) OCT forward imaging probe (green arrow) and RFA catheter (yellow arrow) were strapped side by side for preliminary

evaluation of imaging dynamics due to RF energy delivery. b) The in vivo experiment was conducted in an open chest procedure where the combined OCT and RF catheters were inserted into the right atria. c) Epicardial imaging was also conducted with the combined catheters...... 155

Figure 5. 15. Assessment of tissue contact in vivo. A) Representative OCT image when the catheter was not in contact with the endocardial surface of the right atria. When the OCT catheter was no in contact with the endocardial surface, there was a significant decrease in imaging depth due to blood absorption and scattering. B) Representative OCT image when the catheter was in contact with the endocardial surface of the right atria. Once the catheter was in contact with the tissue, ...... 156

Figure 5. 16. Observance of birefringence dependent bands in vivo in the A) right ventricle epicardium and the B) right atria endocardium ...... 157

Figure 5. 17. Real time, in vivo imaging of a steam pop using Optical Coherence Tomography (OCT). OCT probe and RFA catheter strapped side by side. OCT images taken with forward imaging probe. Observing what we believe to be coagulum building up at the side of the catheter and the development of voids within the myocardium over time...... 158

Figure 6. 1. Clinical significance and utility of monitoring and guidance of ablation therapy using OCT...... 169

Acknowledgements

First, I will like to thank my family for continual encouragement, especially my parents and brother for their unconditional love and support and my grandparents for instilling the love for learning at an early age. Their support and encouragement has been limitless!

I am grateful for mentors and the close friendships I’ve developed over the years that have provided words of inspiration. For all of my committee members, who have provided me with suggestions which have all resulted in the betterment of my dissertation research, and my ability as an engineer conducting interdisciplinary research.

My dissertation research was truly an interdisciplinary project, and I am grateful for the time spent with collaborators: time in the laboratories conducting experiments, teleconferences obtaining feedback, and time in the clinic observing ways in which my research can have a potential impact on health care.

I will like to thank my labmates for providing an environment of intellectual stimulation, their availability whenever I had a question or needed help, and countless conversations over the years.

Lastly, I will like to thank my advisor Dr. Andrew Rollins for giving me the room to grow and feedback to continue developing. His mentorship has spanned many areas including research, teaching and mentoring.

List of Abbreviations

∆λ - bandwidth

2D – Two dimensional

3D – Three dimensional

A-line – axial line

AAD – antiarrhythmic drugs.

AF – Atrial Fibrillation

AV – atrial ventricular

B-scan – cross sectional OCT image

CI – Confidence interval

D – Diffusion coefficient

DTMRI – Diffusion tensor magnetic resonance imaging

EP – Electrophysiology

FDA – Food and Drug Administration

FDOCT – Fourier domain optical coherence tomography

FPS – Frames per second

FWHM – Full width at half-maximum

FT – Fourier transform g – anisotropy coefficient

HIFU – High Intensity Focused Ultrasound

LA – Left atria

LoG – Laplacian of Gaussian

LPF – lines per frame

LV – Left ventricle

µa – absorption coefficient

µs – scattering coefficient

MI – Myocardial infarction

MRI – Magnetic Resonance Imaging

OCT – Optical coherence tomography

OFDI – Optical frequency domain imaging

PBS - Phosphate buffered saline

PSF – Point Spread Function

PSOCT – Polarization sensitive optical coherence tomography

R2 – correlation coefficient

RA – Right atria

RF - Radiofrequency

RFA – Radiofrequency ablation

RV – right ventricle

RVFW – Right ventricular free wall

ROI – Region of interest

SD – Standard deviation

SDOCT – Spectral domain optical coherence tomography

SNR – Signal to Noise Ratio

SSOCT – Swept source optical coherence tomography

SVD – Singular value decomposition

TDOCT – Time domain optical coherence tomography

TTC - Triphenyltetrazolium chloride

VT – Ventricular tachycardia

Characterization of Cardiac Tissue Using Optical Coherence Tomography

Abstract

CHRISTINE P. FLEMING

Radiofrequency ablation (RFA) therapy is the standard of care for the treatment of cardiac arrhythmias. Current techniques to guide ablation therapy utilize low resolution two dimensional fluoroscopic images and functional measurements from the RFA catheter, temperature, impedance and electrograms. High resolution, depth resolved imaging is needed to characterize early structural changes in the myocardium due to disease and therapy. Optical coherence tomography (OCT) is a noninvasive imaging modality that provides high resolution, depth resolved imaging of tissue microstructure in real time. OCT provides subsurface imaging of depths 1-2 mm in cardiac tissue with high spatial resolution (~10 μm) in three dimensions and high sensitivity in vivo. An automated algorithm for fiber orientation quantification in the plane parallel to the wall surface was developed. The algorithm was applied to volumetric image sets of wild type mouse ventricles, and normal and infarctions within rabbit ventricles. Using an ex vivo wedge swine model, we demonstrated that OCT can distinguish necrotic ablation lesions

from untreated tissue and identify precursors to overtreatment. OCT image features were

observed that clearly distinguish untreated myocardium, ablation lesions, epicardial fat, and coronary vessels and assess tissue contact with catheter based imaging, potential

critical structures for real-time guidance of epicardial RFA therapy. A forward scanning

OCT catheter was prototyped that provides contact, cone scanning with no metal was

used to visualization of real time increase in intensity and decrease in gradient strength

and imaging ex vivo of the endocardial surface of the right ventricle submerged in

heparinized blood, where an image of the myocardium was obtained when the catheter

was in direct contact with the tissue, displacing the blood. Using the forward imaging

probe, we demonstrated the first use of OCT to image the endocaridum and visualize

dynamics due to RF energy delivery in a living animal. OCT can provide real-time

direct visualization of RFA treatment to confirm energy delivery and visualize lesion

formation, image in the presence of blood, visualize critical structures and potentially

identify arrhythmogenic substrates. This feedback may increase RFA therapy procedural

success, reduce procedural time, and reduce complication rates.

Chapter 1: Background and Significance

1.1. Catheter Ablation for the Treatment of Cardiac Arrhythmia

Heart disease is the number one cause of deaths in the United States[1]. Cardiac

arrhythmias, abnormal impulse initiation or propagation, afflict millions of patients in the

United States, resulting in frequent hospitalizations and high costs for hospitalizations and medications[1]. Proper action potential morphology and conduction through the heart are required for normal heart function. Current methods for the treatment of arrhythmias include medication, implantable defibrillators and ablation. Medications and devices do not cure arrhythmias but palliate symptoms. Since pharmacological therapies have limited effectiveness, catheter ablation directed at interrupting critical components of arrhythmia circuits has emerged as a prominent approach for the treatment of a broad range of atrial and ventricular tachyarrhythmias. Catheter ablation is particularly attractive because it is the only therapy which offers potential for cure rather than palliation of arrhythmias. Catheter ablation was introduced in 1982 for the treatment of arrhythmia, with laser[2] and direct current (DC)[3, 4] as the first demonstrated energy sources. Catheter ablation using radiofrequency (RF) energy was first demonstrated in vivo in 1987[5]. Additional forms of energy used for catheter ablation of arrhythmias include ultrasound, microwave, and cryoenergy (Table 1. 1).

Table 1. 1 Energy Sources for Catheter Ablation[6]

Energy Source Frequency or Mechanism of Relation of Tissue Contact Advantages Disadvantages Wavelength Heating Heating to Needed Distance (r) from Source Radio 300-700 kHz Resistive 1/r4 Yes Easy, Limited lesion frequency inexpensive, size, charring vast clinical experience Microwave 915-2450MHz Dielectric 1/r2 No Penetrates scar Complex and fat, large catheter design, lesions, linear energy titration catheters possible Laser 300-2000 nm Photon Complex No Large lesions, Difficulty absorption exponential can spare controlling decline Endocardium, depth, complex linear catheters effects with possible tissue properties and distance from source Ultrasound 500 kHz to 200 Mechanical Varies with No Can be focused Difficulty MHz stress and strain focal length for encircling controlling lesions and depth, highly focal lesions far directional from source

1.2. Radiofrequency Ablation (RFA)

Radiofrequency ablation (RFA) is now the standard of care for treatment of many

arrhythmias. RF energy is a form of alternating electrical energy that is used to generate

a lesion in the heart by electrical heating of the myocardium. The goal of catheter

ablation using RF energy is to transfer electro-magnetic energy into thermal energy in the

tissue. Through this process, the area supporting the arrhythmia will be destroyed by

heating it to a lethal temperature. As electrical energy passes through tissue, there is a

voltage drop because the tissue is resistive, resulting in heat is production. RF energy is

delivered in a unipolar form, where the completion of the circuit is by an indifferent

electrode, typically placed on the patient’s skin. Frequencies in the range of 500kHz are

used. The main equations describing the biophysics of RFA are summarized in Table 1.

2. These equations help to interpret key parameters used to monitor RF energy therapy, including power, temperature, and impedance.

Table 1. 2 Biophysics Equations for Radiofrequency Ablation

Voltage VIR= Equation 1

Power PVI= cosα Equation 2 I: total electrical current Current Density ρI 2 Equation 3 Q = R: resistance 4π r2 r: distance from 2 electrode center Heat production per unit ρI Equation 4 H = p: tissue resistivity volume of tissue 16π r2 As shown in equation 4, the distance of the RF electrode and the tissue are critical factors

affecting the amount of heat produced. Because of this, with catheter based RFA,

maintaining stable and good contact with the tissue is critical for heat production in the

tissue and therefore effective formation of a lesion. A narrow rim of tissue (1-2mm) is

heated directly by the RF electrode, zone of resistive heating (Figure 1. 1). The rest of

the lesion forms by passive heating, zone of conductive heating, where the narrow rim acts as a heating source. With continual application of RF energy, a steady state temperature is reached in the tissue. Previous studies have shown that the boundaries of the ablation lesion reach 50oC, where irreversible damage has occurred[6].

Catheter RFA therapy is conducted by inserting flexible catheters into the heart

and where the ablation catheter is in direct contact with the heart surface. Ablation

lesions are generated through a resistive mechanism. Therefore RFA destroys tissue that

triggers abnormal electrical pathways[6], by creating discrete areas of necrosis on the

endocardial surface of the heart which interrupt electrical conduction. Monitoring of

lesion formation is by indirect means from signals at the tip of the catheter, measuring

tissue temperature, impedance, and electrograms. With lesion formation, tissue

temperature rises, impedance decreases, and electrogram signal amplitude decreases.

After lesion formation, electrophysiology tests are conducted to evaluate the effectiveness of the lesion(s).

There are no direct measures of lesion formation currently available. Direct imaging of cardiac tissue during ablation will help guide the precise application of RF

energy and will help monitor the formation of a successful lesion and successful ablation.

Specifically, direct and early detection of ablative lesions would provide clinically

relevant feedback indicating that the target site is indeed undergoing necrosis.

At present, the duration of RFA procedures range from 3 to 8 hours. Moreover,

some procedures, such as RFA for atrial fibrillation are typically associated with the

delivery of dozens of lesions[7-9], producing injury to normal myocardial muscle.

Guidance is also necessary to reduce the number of complications associated with RFA

treatment. Within the 2002 report by the FDA, 95% of ablation procedures are acutely

successful, 90% are chronically successful and 2.5% have major complications. [10] The

FDA defines acute success, chronic success, and complications in the following manner:

• Acute success: non-inducibility of the target arrhythmia

• Chronic success: 3-month freedom from recurrence of target arrhythmia

• Complications: procedure or device related adverse event requiring any

intervention to prevent permanent medical intervention

However, the complications associated with RFA vary depending on arrhythmia targeted.

Complex ablations such as ventricular tachycardia or atrial tachycardia may have complication rates up to 8% (Table 1. 3).[10]

Arrhythmia Acute Success Chronic Success Complications

Atrial Flutter 72 - 100% 85-100% 0 - 6%

Ventricular 66 - 85% 86% 2 - 8% Tachycardia

Atrial Tachycardia 91% 85% 3%

Acceptable >95% >90% <2.5% endpoint Table 1. 3. Success and complication rates of radiofrequency ablation therapy by arrhythmia targeted[10].

A global survey of 777 institutions, and 8745 RFA procedures for the treatment of

atrial fibrillation showed a 76% chronic success and 6% complication rate[11]. However,

when considering procedural success in the absence of anti-arrythmia drugs (AADs),

there was a 52% chronic success, and the other 24% required medication to be

chronically arrhythmia free. It was also noted that medical centers that conduct more

RFA procedures for the treatment of AF have higher chronic success rates[11].

Importantly, the long term complications associated with delivery of numerous

RF lesions are not well reported or studied. Cardiac perforation is one of the most serious complications related to RFA therapy, with a 0.1 to 0.7% incidence[6]. Cardiac perforations have been shown to be associated with high impedance rise.

Thromboembolic events and vascular complications have a 0.6-1.3% incidence rate[6].

In some cases, systemic embolic events occurred without significant impedance or temperature rise or visual coagulum formation. Complications can also occur by inadvertently ablating near or on critical structures such as the AV node or coronary vessels. Although these incidence rates are fairly low, studies have shown that the presence of structural heart disease, multiple ablation targets, and procedural inexperience are additional risk factors. An image-based monitoring system capable of assessing tissue contact, validating lesion formation, identifying critical structures, and discerning early signs of over-treatment may help to increase efficacy and reduce complication rates.

1.2.1. Monitoring and Guidance of RFA

Fluoroscopy, low dosage, real-time X-ray, is the standard imaging tool used to guide RFA therapy. Fluoroscopy is used to navigate the ablation catheter to specific areas within the heart chambers and assess catheter-tissue contact. There are several advanced imaging modalities approaches under investigation to monitor and guide RFA therapy, including magnetic resonance imaging (MRI), [12, 13] computed tomography

(CT) [14, 15], and ultrasound [7, 16].

MRI and CT have been used to obtain the three dimensional (3D) anatomy of the heart for procedure planning and has been recently used for post procedural evaluation.

Three dimensional volumes acquired with MRI or CT can be merged with 3D voltage maps, for procedural guidance targeting arrhythmia substrates. MRI also allows tissue characterization for procedural guidance, such as identification of epicardial fat, fat deposits within the myocardium, pulmonary veins and infarction. Structural information provided by these modalities aid in interpreting electrograms. The use of gadolinium has

been used to increase the contrast of ablation lesions from viable tissue within MRI images [13, 17]. However, MRI is limited because it does not currently provide real-time guidance.

Other than fluoroscopy, echocardiography is currently the only other real-time imaging modality used to monitor and guide RFA therapy. Ultrasound doesn’t have sufficient contrast to visualize ablation lesions without the use of exogenous contrast agents[18].Intracardiac ultrasound has been used to monitor ablation therapy in real time[19] by assessing RFA catheter tissue contact and contact angle[20], visualizing restenosis of pulmonary veins[20], and providing feedback for titration of RF energy to reduce incidence of embolic events due to over-treatment of cardiac tissue[7]. To assess overtreatment, echocardiography imaging relies on the visualization of microbubbles[7], an indirect measure of tissue state. Additionally the formation of microbubbles may be too late of an indication of over-treatment of tissue.

1.2.2. Animal Models for Ablation

Through the development and refinement of RFA procedures, large animal have been used because of their anatomical and electrophysiological similarities to humans. Canine and swine models are well suited for simulating human studies in electrophysiology.

Because of this, canine and swine models have been used to conduct ex vivo experiments to characterize parameters affecting lesion size, and in vivo evaluation of new RF catheters[21, 22] and evaluation of novel RFA procedures[23, 24].

1.2.3. Fiber Orientation

Knowledge of a patient’s heart structure will help to plan procedures, potentially

identifying arrhythmia substrates, critical structures to avoid, and reduce ambiguity when

interpreting electrograms and functional measurements. The structure of the myocardium is important to both electrical conduction and mechanical contractility. Previous research has suggested that heterogeneity in tissue and cellular structure are potential mechanisms for the generation and maintenance of arrhythmias. Due to the anisotropic nature of ventricular myocytes, fiber orientation, the alignment of myocytes within the heart wall, greatly influences the wavefront propagation direction. Abnormal fiber orientation increases the likelihood of arrhythmias.

Over the past forty years, establishing relationships between changes in fiber orientation and normal and abnormal conduction has been an active area of research.

Traditional methods for measuring fiber orientation involve manually measuring angles from slices of histology [25, 26]. It was shown that the fibers change angle in a linear fashion from the epicardium to the endocardium. The use of imaging modalities such as diffusion tensor magnetic resonance imaging (DTMRI) has paved the way to obtain three dimensional fiber orientations from an intact heart. DTMRI is a method that measures the diffusion coefficient of water in a sample to generate three dimensional traces of fibers within a heart[27-29]. A second order diffusion tensor is acquired for each voxel in the volume. The eigenvectors of the diffusion tensor, measured at each location within the volume, gives an ellipsoid where the principle axis corresponds to the direct of the fibers. DTMRI requires long acquisition times to generate a data set with sufficient spatial resolution to resolve fibers. However, fiber orientation within a whole intact heart can be acquired. Fiber orientation has been correlated with functional measurements using optical mapping[30-32]. Optical mapping is a technique that uses fast voltage

sensitive dyes to transducer potential change to an optical signal to allow imaging of

action potentials with high spatial and temporal resolution, in a non-contact means[33].

There is a need for clinical imaging of the myocardial microstructure in vivo to assess structural remodeling during life threatening diseases. At early stages of some diseases healthy myocardium is being progressively replaced by fat or fibrous tissue.

Early detection of this process will help to identify patients at risk of sudden cardiac

death and may serve as an indication for implantable cardioverter defibrillator therapy or

identify targets for radiofrequency ablation. Similarly, basic research of numerous cardiac

diseases would greatly benefit from structural imaging at cellular scale which is possible

only by histology. Early detection requires imaging on the scale of a myocycte, which is

approximately 100 x 10 μm.

1.3. Optical Coherence Tomography

Optical coherence tomography (OCT) is a noninvasive imaging modality that

provides high resolution, depth resolved imaging of tissue microstructure in real time[34,

35]. Images are generated by detecting back reflected light, where contrast is generated by optical index changes in the sample. By measuring singly backscattered light as a function of depth, OCT fills a valuable niche in imaging of tissue microstructure, providing subsurface imaging to depths of 1-3 mm with high spatial resolution (~10 μm)

in three dimensions and high sensitivity (>110 dB) in vivo with no contact needed between the probe and the tissue. With high imaging speeds, high resolution, and fiber based implementations, and functional extensions for measuring tissue birefringence and flow, OCT has made a significant impact in clinical and biomedical applications, including ophthalmology, dermatology, endoscopy and cardiology.

OCT is readily adaptable to a number of clinically relevant imaging applications.

Depth is gated by measuring interference between the sample and a reference using a low

coherence interferometer (Figure 1. 2). Reflection and scattering sites are localized with a

resolution corresponding to the coherence length of the illumination source, lc, given by

2ln2 λ2 Equation 5 l = 0 c π Δλ

Here λ0 is the center wavelength of the source, with full-width at half-maximum

(FWHM) bandwidth Δλ. Hence, low coherence (broadband) sources are employed in

OCT to achieve microstructural imaging approaching the cellular level[36]. A single reflectivity profile as a function of depth is referred to as an A-scan, analogous to

ultrasound A-mode imaging. A two dimensional image (B-scan) is built by collecting many A-scans while scanning the probe beam laterally across the sample (Figure 1. 3-B).

With the use of two scanning mirrors, multiple B-scan images can be acquired by raster scanning the probe beam over the sample surface, resulting in a three dimensional (3D) volumetric image sets (Figure 1. 3C).

Figure 1. 3 OCT image generation of mouse epicardium. (a) 1-D axial scan. (b) 2-D B- scan image generated by transverse scanning and collection of multiple axial scans. (c) 3-

D volume reconstruction from raster scanning across surface, collecting a series of B- scan images.

One of the advantages of OCT is that the axial and lateral resolutions are decoupled. The spatial resolution in the axial direction is equivalent to lc, and the lateral resolution is determined by the focused beam spot size in the tissue. Using conventional optics, there is a tradeoff between the lateral resolution, ∆x, and imaging depth. As shown in, the depth of focus, b, is proportional to the square of the spot size, Equation 6.

Therefore, OCT traditionally uses low numeral aperture lenses to maintain a long depth of focus.

Δx2 Equation 6 b = π 2λ where the spot size

4λ ⎛⎞f Equation 7 Δ=x ⎜⎟ π ⎝⎠d is determined by the focal length f of the focusing optics and beam diameter d.

1.3.1. Time Domain Optical Coherence Tomography

The optical configuration for a fiber based Time Domain, TDOCT system is illustrated schematically in Figure 1. 4.

Figure 1. 4. Time Domain OCT system. Axial scans generated by mechanical scanning of reference mirror. Reflection sites in depth localized to within the coherence length of light source.

As an illustration of generating an axial scan, Figure 1. 4 is a sample with three reflection sites in depth. Light returning from the sample and reference arms is recombined and interferes at the detector. The interference signal is processed and recorded as the reference arm delay line is scanned. Because interference only occurs when the optical path lengths of the sample and reference arms are matched to within the coherence length of the light source, reflection and scattering sites are localized within a resolution corresponding to the coherence length (Figure 1. 4-C). By monitoring the envelope of the detected interferometric pattern (i.e., photodiode current) as a function of the reference arm delay, a profile of sample reflectivity versus depth is obtained (Figure

1. 4-D) [37].

ρ PRs s In the shot noise limit, the signal to noise ration (SNR) is , where Ps is the 2eB power incident on the sample, Rs is the power reflectivity of the sample, e is the electronic charge, B is the detection bandwidth, and ρ is the detector responsivity given by ρ = ηλ0 hce . Here η is the detector quantum efficiency, λ0 is the optical source center wavelength, h is Planck’s constant, and c is the free space speed of light.

1.3.2. Fourier Domain Optical Coherence Tomography

The principle of Fourier domain OCT (FDOCT) is capturing a spectral interference pattern[38, 39] instead of the temporal interference pattern. Parallel detection in FD-OCT increases the integration time at each A-scan and thus improve the sensitivity of FD-OCT about 20dB over TD-OCT without a trade-off in the imaging

ρPRΔ t speed[40, 41]. The SNR for a FDOCT system, assuming shot noise limited, is ss . 2e

The SNR depends on the source power, Ps, the sample arm reflectivity, Rs, the detector responsivity, ρ , and the integration time of the camera or sweep time of the swept source, ∆t. The increase in sensitivity of FD-OCT systems allows increased A-line rates, enabling video rate imaging and acquisition of 3D image sets in short time periods. For many clinical applications, high imaging speed is critical to reduce motion artifacts and observe dynamics.

Figure 1. 5. Fourier Domain OCT implementations. (a) Spectral Domain OCT (SDOCT). (b) Swept Source OCT (SSOCT) or Optical Frequency Domain Imaging (OFDI)

Fourier domain setups can be implemented using a broad band light source and spectrometer in the detector arm of the interferometer, called spectral domain OCT (SD-

OCT), Figure 1. 5A, or using a single detector in conjunction with a swept-frequency optical source, called swept source (SS-OCT) or optical frequency domain imaging

(OFDI), Figure 1. 5-B. In FDOCT the reference arm is stationary and the captured spectral interferogram represents the inverse Fourier transform of the backscatter profile as function of depth, or OCT A-scan (Figure 1. 5). The recorded FDOCT signal must be

Fourier transformed to generate an OCT image. The photo current detector as a function of wavenmber, known as the spectral interferogram, is composed of three terms

(Equation 8). The first term is a constant DC offset. The second term is composed of a sum of sinusoidal terms, where each cosine is proportional to the square root of sample reflectivity. This is the signal of interest. The depth of the scattering event is encoded in the frequency of the sinusoidal term. The third term is composed of a sum of auto correlation terms, due to mutual interference from each reflection site in the sample

IkD ()= DCACMI++ ρ ⎡⎤⎛⎞N DC=+⎢⎥ S() k⎜⎟ RRSn∑ R 4 ⎣⎦⎝⎠n=1 Equation 8 ρ ⎡ N ⎤ AC=−⎢ S() k∑ RRSn Rcos() 2 k() zR z Sn ⎥ 2 ⎣ n=1 ⎦ ρ ⎡ N ⎤ MI=−⎢ S() k∑ RSn R Sm cos() 2 k() zSn z Sm ⎥ 4 ⎣ nm≠=1 ⎦

where S(k) is the wavenumber spectrum, RR is the reflectivity of the reference mirror, RSn is the reflectivity of the nth reflection site in the sample, zR is the position of the reference mirror, zSn is the position of the nth reflection site in the sample, and ρ is the detection sensitivity.

First step taken after obtaining the spectral interferogram, is to remove the DC component of the signal. This is accomplished by blocking the sample arm and capturing a frame to subtract from all subsequent frames or averaging axial scans average the A- scans of the first image to obtain the DC spectrum. The next step is to resample the spectrum so that it is linearly spaced in k. The resulting signal is Fourier transformed, transforming from k to z, to obtain the axial scan.

An example obtaining an axial scan with a FOCT system of a sample with three reflection sites is shown in Figure 1. 6. An axial scan is generated by taking the Fourier transform of the spectra interferogram. Since the spectral interferogram is real, the

Fourier transform is an even function. This results in a mirrored image on the opposite side of zero path length. Samples that are equidistant from the zero path length cannot be resolved becausecos( 2kzΔ=) cos( −Δ 2 kz) . This is called the complex conjugate artifact. If the entire sample being imaged is located entirely on one side of zero path length, this complex conjugate artifact is not serious and only the positive or negative distances are displayed. However, if a portion of the sample, is located above the zero path length position, the mirrored image will overlap and this cannot be removed with standard imaging processing. Several techniques have been proposed to solve this problem[42-47].

The imaging range of a FDOCT system is fixed based on the optical and detection design. The frequency of the modulation of the spectral interferogram increases as the reflector site increases in distance from the zero path length. With a finite number of sample points, N, the frequencies that can be resolved is limited by the Nyquist theorem.

As shown in Equation 9, the imaging range, D, is a function of the center wavelength, λc, the spectral range covered by the spectrometer, ∆λ and the number of pixels on the detector. If the complex conjugate ambiguity is resolved, N 4 is replaced by N 2 .

1 λ 2 Equation 9 Δ=DNc 4 Δ λ Within FDOCT systems, the sensitivity degrades with increasing distance from the zero path length match[48, 49]. This sensitivity fall off is related to the spectrometer optics and the pixel width of the detection cameras in SDOCT or the instantaneous linewidth of the swept light source. The theoretical -6dB fall off can be calculated using

Equation 10, whereδrk is the spectral resolution.

ln2 Equation 10 Δ=z−6dB πδrk

1.3.2.1. Spectrometers

Spectrometers are a key component in the development of SDOCT systems. The spectral interferogram recorded by the spectrometer must be Fourier transformed to generate an OCT image. Spectrometers used for FDOCT consist of a diffractive grating and collimating and objective optics, and a line-scan camera. The diffraction angle of light dispersed by the grating is a nonlinear function of wavenumber k=2π/λ. Therefore, the spectrum recorded by the line-scan camera is unevenly spaced in k. However, a spectral interferogram that is a linear function of k is needed to inverse FT the data into the spatial domain. Current practice is to interpolate the nonlinear spectral interferogram and rescale the data into the wavenumber domain prior to the FT. Our lab recently introduced a spectrometer with dispersion that is linear in wavenumber (linear k) for

FDOCT imaging and demonstrated improvement in falloff and computing time as compared to an equivalent conventional spectrometer[50].

The other key element in spectrometers is the detector. The imaging range is directly proportional to the number of detecting elements in the line scan camera. As an example, with the current SDOCT setup at CWRU, our spectral range is 100nm, center wavelength 1310nm, and detector with a 1024 element PDA line scan camera from

Goodrich. This results in an imaging range of 4.3mm.

1.3.3. Polarization Sensitive OCT

Polarization-sensitive OCT (PSOCT) provides information on the polarization state of the light reflected from the samples under study[51]. Changes in the polarization state of light are dominated by two mechanisms: scattering and birefringence. Scattering

changes the polarization state of light in a random manner. Birefringence is a material property exhibited in highly organized tissue such as collagen, where anisoptropic indices of refraction are observed for varying polarization states. Beyond structural imaging,

PSOCT provides additional contrast to identify organized tissue architecture and abnormal or damaged tissue. PSOCT has been used to evaluate collagen content within intravascular plaques[52-54], normal fiber organization within the myocardium of animal models[55], tissue damage due to ablation therapy[56, 57], infarction[55], and the nerve fiber layer within the retina[58-61]. When imaging through fiber catheters, the rotation of the fiber may cause stress-induced birefringence, which can adversely affect PSOCT measurements. Through the use of frequency multiplexing, an OFDI system has been developed that allows catheter-based PSOCT independent of the fiber birefringence [62].

1.3.4. OCT Catheter Probes

To translate OCT imaging technology to practical application, application- specific scanners must be developed to deliver probe light to the sample of interest.

Catheter probe development has been used extensively for gastroentestinal endoscopy and intravascular imaging. A variety of designs have been employed including forward imaging[63-66], sideviewing probes[67], both forward and sideviewing probes[68], spectrally encoded probes[69], and catheters based on MEMs actuated mirrors[70, 71].

Higher image acquisition speeds with FDOCT technology has enabled large volumes of internal lumens to be imaged. Probes are developed with a specialized shaft, which is axially flexible and torsionally rigid, mechanically supports the optical elements of the probe. The probe optics are designed to minimize loss and back reflection of probe light.

The probe sheath materials are selected for biocompatibility, compatibility with standard

cleaning and sterilization procedures, and optical and mechanical properties. To take advantage of the technology developments in high speed imaging and broadband light sources, recent catheter probes have been developed with both rotational and pullback capabilities for volumetric imaging of internal lumens. Specialized optical designs have been developed to maintain a circular spot profile and reduce astigmatism and accommodate broad bandwidth light sources for high axial resolution[72]

1.3.5. OCT Imaging of the Cardiovascular System

Cardiovascular disease is the leading cause of morbidity and mortality in the

United States[1]. Imaging has played a vital role for early diagnosis of cardiovascular diseases, monitoring and guidance of procedures, and characterization of preclinical models of disease. The unique features of OCT have made it a powerful tool for cardiovascular imaging, from basic scientific research to clinical applications. In particular, cardiovascular OCT is a technology that can potentially become standard for the detection and treatment of atherosclerotic plaques. Postmortem studies of patients who experienced a myocardial infarction have identified common features of plaques including the presence of large lipid, a necrotic core, a thin fibrous cap (<65 micron), microcalcification and inflammatory cells[73]. Resolution of conventional imaging modalities has limited the ability to visualize these features in vivo. With resolution on the order of 10µm, and imaging depth of approximately 2mm, OCT can visualize thin fibrous cap [74] and accurately differentiate between major components of the atherosclerotic plaque (fibrotic, lipid, calcium and collagen)[52, 75, 76]. Preliminary in vitro studies show the promise of automated differentiation of plaque components using a single scattering model to extract the optical attenuation coefficient[77, 78]. OCT has

also demonstrated the ability to visualize and quantify macrophage density in atherosclerotic plaques[79-81]. In addition to evaluating plaque composition, OCT shows great promise for assessing vascular response to stents[82-85]. OCT has been used to assess stent apposition, coverage, and presence of thrombosis.

Broader adoption of OCT in the clinical setting has been limited by the need of temporary vessel occlusion and saline flushing to provide an optically clear field of view during image acquisitions. With the introduction of high speed FDOCT systems, these limitations are disappearing. An optically clear field of view is still necessary. However, with a single 4 second saline flush without vessel occlusion, volumetric imaging of an entire coronary segment is possible. This has been shown in vivo in large animals and humans[86]

Recent research has demonstrated that OCT has great potential for studying the cardiovascular system within small animal models. The structure of the myocardium is important to both electrical conduction and mechanical contractility. OCT has been demonstrated to visualize critical structures related to electrical conduction, including the purkinje network[87], and imaging the fast and slow pathways in the atrial-ventricular

(AV) node[31, 88], myofiber organization[31, 89] in animal models and in vitro preparations of human tissue[90]. In 2002, Villard et. al. showed real-time imaging of ventricular wall thickening over the course of several cardiac cycles within six murine adult hearts[91]. The use of a blood substitute was used to eliminate the effects of light absorption due to presence of blood.

Due to its high spatial and temporal resolution, imaging penetration depth, and non-contact nature, OCT fills a valuable need for imaging the structure and function of

the developing cardiovascular system. OCT has been used to image the embryonic chick heart over the first few days of development, where the heart transitions from a tube to a four chamber heart. OCT has promise to be a tool to enable studies of normal and abnormal heart development and to evaluate surgical or pharmacological interventions.

Three dimensional OCT imaging of fixed[92] or living[93-96] embryonic hearts allows for analysis of morphological and blood flow[97-102] changes as a function of genetic deficiencies[92] or environmental perturbations[101].

High speed image acquisition has allowed visualization of the beating avian and murine hearts in three dimensions Longitudinal imaging of embryonic heart development has been demonstrated by integrating the sample arm into a microscope inside a custom built incubator. The incubator maintains oxygen, temperature, and humidity levels which allows five dimensional imaging of avian heart development[103].

1.3.6. Image analysis OCT, as with other coherent imaging modalities, is susceptible to speckle noise.

Speckle in OCT degrades visual image quality and can interfere with computer image analysis. Approaches have been taken including angle compounding[104-106], frequency compounding[107] and post-processing image analysis to reduce speckle within OCT images[108]. Post-processing approaches to reduce speckle have been demonstrated using averaging[109], median[109, 110] and Wiener[89] filtering kernels, and adaptive filters such as the rotating kernel transformation method[109, 111], spatially adaptive two dimensional wavelet filter[112], fuzzing thresholding in the wavelet domain[113], Lee and adaptive Wiener filters[114]. Figures of merit to evaluate the performance of noise reduction techniques have included contrast to noise ratio[112, 115], signal to noise ratio[112, 113, 115], equivalent number of looks[112, 113, 115], edge preservation[113],

and computational time[113]. In addition, Gargesha evaluated the performance of the median filter, Wiener filter, orthogonal wavelet and non-orthogonal wavelet filters on four-dimensional image sets of embryonic hearts through volume rendering and semi- automatic segmentation[116] and Rogowska evaluated the performance of mean, median and the rotating kernel methods for improving edge detection within OCT images of cartilage[109]. The choice of noise reduction filter is application specific, where computational time and improvement in figures of merit are tradeoffs.

OCT images can be described by an image convolved with a two dimensional point spread function (PSF), resulting in a degraded, blurred image. Using prior information about the imaging optics and optical properties of the imaged tissue to derive the system PSF, iterative deconvolution has been applied to the image to increase resolution[117-122], by deconvolving the original image from the degraded image.

Quantitative evaluation of images may provide additional information on the tissue pathology. These algorithms generally require at least three steps. The first is segmentation to identify regions of interest (ROI) in the image. The second is feature extraction, computation on the image pixels resulting in a quantitative value representing a feature of the image. The third step is classification, where feature values are used to make a decision about the composition of the tissue represented by the image. Currently, image segmentation in OCT has been largely limited to manual segmentation and thresholding[123]. With the rise of Fourier domain optical coherence tomography, image data sets have dramatically increased in size, and there is a great need for automated segmentation and region of interest (ROI) identification[124]. The fundamental basis for image segmentation and ROI identification is detection of edges. Automated edge-based

segmentation techniques have been applied for detecting bone-cartilage borders[109], a combined edge-based and morphological processing has been used to identify colonic crypts within en face images[110], a Markov boundary model[125] and Perona-Malik filter[126] to detect retinal layers, and intensity variations[127] and graph based segmentation of surfaces within three dimensional SDOCT datasets[128] to identify the macula. Edge based methods use local information, resulting in a degraded performance in areas with low signal-to-noise.

Numerous groups are working towards developing efficient and robust tissue classification algorithms to extract physiological data from acquired OCT images.

Automated algorithms developed to identify differences in optical properties and image texture for tissue classification can aid in procedural guidance and evaluation. In intravascular OCT, optical attenuation coefficient of atherosclerotic plaque has been locally measured by quantitative OCT[77, 78]. Optical properties such as attenuation, backscattering, scattering, and anisotropy coefficients can be obtained from OCT images using image analysis. There are a variety of models for OCT signals including single scattering models[129], multiple scattering models[130] and models for multiple layered samples[131]. These models have been used to analyze the optical properties of plaques

[132], neuronal cells [133] and weekly scattering medium [134] for the purpose of tissue classification. Recently, an improved quantification by combining measurements of optical backscattering and attenuation coefficients has been demonstrated for further enhancing this differentiation[135]. Quantitative OCT has demonstrated quantification of macrophage density in atherosclerotic plaques, where large variance of OCT signal can be attributed to the heterogeneous structure induced by a high density of

macrophages[79, 80]. Utilizing additional contrast from functional OCT spectroscopic

OCT[136]. Robust analysis tools utilizing intensity based and functional OCT are needed to overcome obstacles present for automated analysis for in vivo use including speckle noise, motion artifacts and residual blood.

1.4. Objective

There are limited tools available to image myocardial tissue microstructure at high temporal speeds, high resolution, within intact preparations. The objective of this dissertation is to develop Optical Coherence Tomography (OCT) technology for the use of myocardial tissue characterization. New image processing algorithms and OCT scanners were developed for two applications: 1) imaging and quantifying myofiber orientation within intact heart preparations and 2) monitoring and guidance of cardiac radiofrequency ablation therapy. There is a need for clinical imaging of the myocardial microstructure in vivo to assess structural remodeling during life threatening diseases.

Early detection requires imaging on the scale of a myocycte, which is approximately 100 x 10 μm. Chapter 2 will summarize an algorithm developed to quantify fiber orientation within three dimensional OCT image sets. We will also explore the use of OCT to image various animal models and characterize myocardial infarction. The current implementation of RFA in clinical practice is associated with significant unmet technological need. There are no direct measures of the successful delivery of ablation lesions. Technology for directly monitoring ablation lesion formation during procedures in real-time may decrease procedural time and improve patient and operator safety.

Within Chapter 3, we will demonstrate that there is a signal present within OCT images to distinguish RFA lesions from untreated tissue. Chapter 4 will explore the use of OCT

for tissue characterization, for potential guidance of epicardial RFA using OCT. Chapter

5 will translate the findings in Chapters 3 and 4 for in vivo use, through the development of a forward imaging catheter. Chapter 6 will include a summary of the dissertation, clinical implications, and future technology development and experimental proposals.

Chapter 2: Characterization of Arrhythmogenic Substrates Using Optical Coherence Tomography

In the previous chapters, we have demonstrated that OCT can validate lesion formation, distinguish necrotic tissue from untreated tissue, visualize dynamics due to lesion formation with catheter based imaging, and visualize critical epicardial structures for procedural guidance. In this chapter, we explore the use of OCT to visualize arrhythmogenic substrates and potential targets for ablation.

2.1. Introduction

The structure of the myocardium is important to both electrical conduction and mechanical contractility. Due to the anisotropic nature of myocytes, cell-to-cell coupling and fiber orientation (alignment of myocytes) directly influence the wavefront propagation, force generation and efficiency of contraction. Abnormal fiber orientation caused by a disease such as infarction or hypertrophic cardiomyopathy is a substrate for arrhythmia that may result in sudden cardiac death.

Early detection of this process will help to identify patients at risk of sudden cardiac death and may serve as an indication for implantable cardioverter defibrillator therapy.

Similarly, basic research of numerous cardiac diseases would greatly benefit from structural imaging at cellular scale which is possible only by histology. Early detection requires imaging on the scale of a myocycte, which is approximately 100 x 10 μm.

Proper action potential morphology and conduction through the heart are required for normal heart function. Due to the anisotropic nature of ventricular myocytes, fiber orientation, the alignment of myocytes within the heart wall, greatly influences wavefront propagation. Therefore, abnormal fiber orientation increases the likelihood of abnormal cardiac rhythms, or arrhythmias.[137]

Current methods for measuring fiber orientation range from histology [25, 26] to diffusion tensor magnetic resonance imaging (DTMRI) [138]. It is known that fiber angle varies in a nearly linear fashion from the epicardium to the endocardium [25, 26]. Karlon and colleagues developed an automated image analysis method for measuring fiber orientation from histology slices[139]. The method uses intensity based gradients within an image to calculate fiber orientation and angular standard deviation within two dimensional histology slices of mouse hearts.

Heterogeneity in cardiac tissue microstructure is a potential mechanism for the generation and maintenance of arrhythmias. Abnormal changes in fiber orientation or fiber disarray increase the likelihood of arrhythmia. We present and validate an automated method for quantifying fiber orientation from structurally intact excised cardiac tissue preparations using intensity-based gradients applied to OCT en-face images.

2.2. Two dimensional fiber orientation algorithm

Three dimensional optical coherence tomography image sets of the endocardial surface of an isolated right ventricular free wall (RVFW) preparation from a New

Zealand White rabbit were used in this study. The protocol was approved by the

Institutional Animal Care and Use Committee at Washington University. The experimental procedures have been previously described [140]. Briefly, the rabbit was anesthetized intravenously and the heart was removed and coronary perfused with Tyrode solution on a Langendorff apparatus. The heart was removed from the Langendorff apparatus and placed in a bath of ice-cold Tyrode solution. The right ventricular free wall was dissected, stretched, and pinned epicardial side down onto a silicon disk. The sample was placed in 3.7% formaldehyde for one day and 20% sucrose solution for an additional two days. This dehydration step improves the visibility of the fibers under OCT imaging.

A microscope based OCT system [141] was used to image volumes of the sample. The axial and lateral resolution of the system was approximately 10 micrometers

(in air). The three datasets presented in this work vary in structural complexity (Figure

2. 1A). Each volume was 4.5mm x 4.5mm x 3.16mm, corresponding to a pixel size of

11.25μm x 11.25 μm x 4.86 μm. To calculate fiber orientation from two-dimensional

OCT images, a modification of the intensity based gradient algorithm described by

Karlon et. al. [139] was used.

Figure 2. 1: 3D OCT data sets of the Right Ventricular Free Wall (RVFW). A) Three- dimensional OCT images of the RVFW, data sets A1) RVFW1, A2) RVFW2, A3) RVFW3. Arrow in A1 points to cross sectional (en-face) slices. B) View of three en- face slices in depth within 3D OCT data sets. Fiber structure is visible within en-face slices. The uneven sample surface (B1) and trabeculations (B2 and B3) induce low frequency changes and produce shadows within subsequent en-face slices, inducing a gradient that may be greater than the gradient produced by the fibers.

Within en-face OCT images (Figure 2. 1B), uneven sample surface topology (e.g.

RVFW1) and shadows cast by trabeculations (e.g. RVFW2 and RVFW3) cause low

spatial frequency changes in the background intensity within en face images. These artifacts introduce unwanted intensity gradients within the en face image. A two dimensional second order Butterworth high-pass filter was used under the assumption that the surface topology and shadowing artifacts have lower spatial frequency components compared to visible fiber structures. The high-pass filter was convolved with the en-face OCT image to suppress variations in background intensity. The high- pass filtered en-face OCT image was convolved with a Wiener filter for noise reduction.

Two-dimensional 3x3 Sobel filters were used to estimate local gradients in the image. Gx and Gy are defined as the convolution of the horizontal and vertical Sobel filters with the 2D en-face filtered OCT images. For each pixel, the magnitude of the

22 gradient, G(i,j)= GGx + y and the gradient direction, θ’ = atan()GGyx, was calculated.

Within a small local window of the image, W, the dominant local direction of the

W gradient was computed by taking the maximum of the angular distribution function, Aθ , a function of G(i,j) and θ’(i,j). The angular distribution function is a fit of a radial normal distribution to the distribution of angles within the local window.

W exp( 2cos(θθ− '(ij , )) AGijθ = ∑ (, ) (,ijW ) exp(2) where : 0oo≤≤θ 180

22 Gi(, j )=+ Gxy (, i j ) G (, i j )

θ '(ij , )= a tan( Gyx / G )

The directions of the cardiac fibers were assigned as perpendicular to the direction of the dominant local gradients. After fiber orientations were assigned to each window within the image, two criteria were used to reject invalid fiber orientation assignments.

First, the algorithm identified windows with no tissue present by using a threshold on the average pixel intensity within the window. Secondly, to measure the confidence in fiber orientation measurements produced by the automated algorithm, a D’Agostino-Pearson’s

κ2 (normality) test was conducted on the angular distribution of calculated orientations within each window. High values of κ2 indicate that the angular distribution function is not a normal distribution, and therefore, the confidence in the fiber orientation

assignment in that window is low. For example, a non-normal distribution may occur in windows where the image is too noisy or contains multiple fiber directions. Threshold values for κ2 (0.02) and average intensity values within a window (80, 1.5 times the noise floor) were selected. The automated algorithm was implemented using the software package Matlab 7.3.0.267 R2006b (Copyright 1984-2006, The Mathworks, Inc.).

In order to validate the method quantitatively, an investigator blinded to the results of the automated algorithm manually measured fiber orientation angles on the same en-face OCT images that were analyzed by the automated algorithm. En-face images were analyzed in increments of 25 μm in depth for all three data sets both manually and using the automated algorithm. Results from the automated algorithm were compared to manual measurements by analyzing the mean and standard deviation of fiber orientation assignments for each depth and orientation as a function of depth. This comparison was made using several window sizes, but the mean of the absolute difference was the lowest for a window size of 563 μm x 563 μm (almost four myocytes in length, assuming that an adult cardiac myocyte has an average length of 150µm).

Results using a 563 μm x 563 μm windows were resampled to obtain 256 vectors per image. Under these settings, the two dimensional fiber orientation algorithm runs in less than 7 seconds per image using a Dell Precision Workstation 390 with a Dual-Core processor running at 2.66GHz and 2GB SDRAM memory, 533MHz with Windows XP

Professional.

2.2.1. Results

Accurate fiber orientation measurements were obtained from all three data sets up to 500µm below the sample surface. Figure 2. 3 shows example vector plots of fiber orientations overlaid on raw en face OCT images. The low frequency background intensity change and shadows created by endocardial trabeculations are apparent within the raw OCT image. Preprocessing of the data effectively reduced the gradient contribution of these features, producing accurate measurements of fiber orientation for samples of varying structural complexity.

Figure 2. 3: Fiber orientation in the rabbit right ventricle. Automated algorithm enables quantification of fiber orientation in the plane parallel to the wall surface for samples of varying structural complexities. A) RVFW1, B) RVFW2, C) RVFW3. Although there are gradients within the images due to uneven surfaces (A) and shadows created by trabeculations (B, C) the two-step filtering process effectively reduce their contribution to the gradient magnitude calculations.

It has been well established that normal fiber orientation varies nearly monotonically from the epicardium to the endocardium. Figure 2. 4 shows fiber orientation measurements as a function of depth for representative 281 μm x 281 μm x

500 μm volumes from each of the three data sets. In all three cases, there is a nearly linear change in fiber orientation with depth. Comparing the slopes for the RVFW1 volume shows that the automated measurements correspond very well to the manual measurements (Figure 2. 4A). A quantitative comparison of fiber orientation

assignments was conducted on an 1125 x 1125 x 500 µm volumes randomly selected within the RVFW1 data set (Figure 2. 4A-1). Fiber orientation assignments made by the automated algorithm correlated very well to manual measurements, with a 1.002 slope and a 0.823 correlation coefficient.

Lethal arrhythmias commonly arise from structural heart disease, such as following myocardial infarction (MI). The epicardial border zone is characterized by reduced conduction velocity and increased anisotropy associated with electrical reentry and sustained ventricular arrhythmias.[142-146] Treatment of ventricular tachycardia by cardiac RFA involves ablating small islands of viable tissue within the border zone of the infarction. Optical guidance has the potential to identify patches of viable fibers to be

ablated to terminate the arrhythmia.

Preliminary experiments were conducted to evaluate if OCT can be used to assess tissue structure in an infarction model. The healed myocardial infarction rabbit arrhythmia model developed by Yuanna Cheng was used in this study[147].Briefly, in vivo survival surgery was used to create a discrete myocardial infarction in the left ventricle (LV) of the rabbit heart by ligation the branches of the left circumflex artery. In vivo survival surgery was conducted at the Cleveland Clinic and OCT imaging was conducted at Washington University in St. Louis. The rabbit was anesthetized intravenously and the heart was removed and coronary perfused with Tyrode;s solution on a Langendorff preparation, fixed in 3.7% formaldehyde and subsequently placed in

20% sucrose solution two days. A microscope based 1310nm TDOCT system [141] was used to image volumes of the sample. Each volume was 4mm x 4mm x 3.16mm. Three dimensional images were taken of normal, infarction and border zone. There was a transition from regular fiber orientation to disarray to loss of fibers within the infarction

(Figure 2. 5).

Although OCT has limited penetration depth into myocardial tissue of 1-2 millimeters, this is enough to image through the majority of the ventricular wall of a mouse heart. Three dimensional image sets were acquired of three mice hearts with

healed myocardial infarctions and one control heart with the microscope integrated

SDOCT 1310nm system. The samples were formalin fixed and imaged intact. As shown in Figure 2. 6, significant ventricular thinning, increased signal intensity, and loss of birefringence banding was observed within the area of infarction. However, fibers were not visible within image slices parallel to the sample surface.

2.2.3. High Resolution Imaging

With a standard time domain system, OCT has demonstrated imaging and quantification of fiber orientation in the plane parallel to the wall surface. During these experiments, the right ventricular free wall of a New Zealand White rabbit was dissected, stretched, and pinned epicardial side down onto a silicon disk. This procedure ensured that the sample surface was relatively flat. An optical clearing procedure was conducted to enhance the visibility of the fibers within the OCT image sets. These are procedures that cannot translate directly to in vivo imaging. Therefore, developing an OCT scanner

with high axial and transverse resolution may alleviate the need for optical clearing and fixation protocols.

The average adult ventricular myocyte varies with species. For example, a rat myocyte has a depth (μm), width (μm) and length (μm) of 5, 25, and 150[148]. To adequately image fiber orientation using OCT, high resolution is needed in both the axial and transverse directions. Previous studies in our lab has shown that standard OCT imaging systems can be used to measure transverse fiber orientation angle in a fixed rabbit right ventricular free wall preparation[89]. However, standard OCT scanners have axial and transverse resolution on the order of the width of a myocyte. Since the smallest dimension of the myocyte is the depth, high axial resolution provided by broad band light sources will greatly enhance the ability for optical coherence tomography to image both fiber orientation inclination and transverse angles. A high resolution OCT system provides high axial resolution. Therefore samples can be imaged without stretching and pinning the sample and without an optical clearing procedure.

Three dimensional OCT image sets of the epicardial surface of an intact heart from a mouse heart were taken with a high resolution FDOCT system with a light source centered at 820nm. The mouse was anesthetized and the heart was removed and coronary perfused with Krebs solution and then perfused with formalin on a Langendorff apparatus. The protocol was approved by the Institutional Animal Care and Use

Committee at Case Western Reserve University. The heart was stored in formalin for one day until imaging. Spectral inteferograms were recorded onto a 4096 pixel line scan camera (Atmel) with a 14 kHz line scan rate. The system has a 95 dB SNR, 1mm -6dB fall off, and 4 and 7 micrometer axial and lateral resolution respectively. The image area

was 2x2 mm. Three wild type mouse hearts and one New Zealand White rabbit heart was imaged with the high resolution system. Presented in Figure 2. 7 are fiber orientation measurements obtained from an intact wild type mouse heart. Within the first layers, fiber orientation is visible.

Figure 2. 7: High resolution imaging of the mouse myocardium. A) Three dimensional reconstruction of wild type mouse right ventricle and right atria. B,C) Representative B- scans from volume. D) Slice parallel to epicardial surface, with overlay of fiber orientation vectors

For future applications, a high resolution 1310 nm OCT will provide both high axial resolution and increased depth penetration. Maintaining a relatively constant transverse resolution with depth will increase the visibility of fibers in depth. In addition, there is a need to extend this algorithm to quantify fiber orientation in three dimensions to accommodate samples with irregular fiber orientation patterns and samples with curved surfaces.

2.3. Analysis of an Optical Mapping Inverse Model

Optical mapping is a technique that uses fast voltage sensitive dyes that transduces membrane potential change to an optical signal[33]. This allows imaging of action potentials with high spatial and temporal resolution, in a non-contact means. A two dimensional time dependent image is generated, where action potential morphology and conduction can be further analyzed. Due to the electrical connection within the heart and light propagation, an optical mapping signal can integrate 300-2000 μm of depth into a surface map, where the contribution to the signal as a function of depth is unknown. The problem of understanding the depth heterogeneity of action potential morphology and conduction is complex due to the spatial and temporal variability in action potential characteristics. Plunge microelectrodes and optical probes[149, 150] can track potential changes with high temporal fidelity, but lack the spatial resolution necessary to characterize the conduction pattern. Numerical simulations[151-155] of fluorescence propagation and of ionic conduction have also been used to gain insight into the three dimensional nature of action potential conduction and morphology.

The optical mapping inverse problem is solving for the kinetics parameters that describe the three dimensional potential within a sample that reproduces the two

dimensional surface optical mapping signal. The flow diagram for the inverse model is shown in Figure 2. 8.

The steady state diffusion equation for highly scattering media was used to model both the excitation and emissions light. Where the source term in the diffusion equation for the emissions light is a term that is proportional to the excitation light and the membrane potential at a specific location in the tissue. The membrane potential distribution within the sample was model with the Beeler Reuter Model for ventricular

fibers[156]. This model was chosen for its simplicity. However, it maintains all of the characteristics of a ventricular action potential. The Beeler Reuter model solves a system of eight, first order differential equations. Six Hodkin Huxley kinetic parameters were used to describe the current flow, an equation was used to describe the kinetics of the intracellular calcium concentration, and an equation to describe the membrane potential.

The modeled outlined is a nonlinear least squares problem. The objective of the nonlinear least square model is to iteratively solve for kinetic parameters that describe the three dimensional distribution of membrane potential that can give rise to a similar optical mapping signal similar to the recorded optical mapping signal. The word similar refers to the stopping condition of this iterative procedure of updated kinetic parameters.

The equations used to describe optical mapping signal are nonlinear function of the kinetic membrane potential parameters. Many nonlinear problems are solved by iteratively solving linear least square problems. Therefore, a detailed analysis was carried out to evaluate which methods are appropriate to solve the inverse problem. This included 1) properly formulating the problem, 2) computing a singular value decomposition of the problem matrix, and 3) evaluating the conditioning and stability of the problem. Methods to enable solving the optical mapping inverse problem with a linear least square solution such as singular value decomposition truncation and regularization of the linear least square problem were evaluated.

The Jacobian matrix is heavily used to solve nonlinear least squares problems.

The Jacobian is defined as the partial derivative of a function with respect to a set of variables. This results in an mxn matrix where m is the number of outputs and n is the number of parameters. The Jacobian for the optical mapping inverse model is the derivative of the computed optical signal with respect to the kinetic parameters describing the action potential. The only portion of the membrane potential that is a function of the kinetic parameters is the current terms. Therefore, we have an analytical representation of the Jacobian.

∂Vopt ∂ ⎛⎞ϕϕ(zz=−01) ( =) Jacobian==−⎜⎟ D ∂∂ppii⎝⎠ dz −∂D ⎛⎞ ∂ ==−=⎜⎟ϕϕ()zz01() dz⎝⎠∂∂ pii p −∂D ⎛⎞ ∂ ==−=⎜⎟Jzion ()01 Jzion () dz⎝⎠∂∂ pii p 256x 12 Jacobian ∈

The Beeler Reuter model for ventricular action potentials has twelve conductance and kinetics parameters to describe the action potential. This resulted in a Jacobian of the size tx12, where t is the length of the time series being analyzed. It was observed that the

Jacobian was not full rank. The rank of a matrix gives an indication of the linear independence of the columns within the matrix. After a few iterations, the number of parameters was reduced until the Jacobian had full rank. This occurred when the number of parameters was reduced to six.

The first class of methods used to solve nonlinear least square problems involves solving an affine model of the residual around xc, the value of parameters at the current step. The residual is defined as the difference between the computed optical mapping signal from the measured optical mapping signal. An initial guess of the conductance parameters is used to start the algorithm, and an updated set of parameters, x+, were computed using the following equation using the Gauss Newton method.

−1 TT x+ =−xJxJxJxRxccccc( () ()) () ()

To compute x+, the equation can be rearranged in the standard form of Ax=b.

T −1 x+ =− xccccc( Jx() Jx ()) Jx()() Rx T ()Jx()cc Jx ()()()() x+ −=− x c Jx cc Rx TT Jx()cc Jx () x+ =− Jx () ccccc Jx () x Jx ()() Rx

Ax11= b 1 T AJxJx1 = ()cc ()

xx1 = + TT bJxJxxJxRx1 =−()()ccccc () () ()

The Damped Gauss Newton method is a modification of the Gauss Newton method with a line search. This enables the next iteration to go in the direction of steepest decent. This improves the ability of the algorithm to work with problems that are sufficiently nonlinear and problems with large residuals. Below is a transformation of the

Damped Gauss Newton method into the common Ax=b form.

T −1 x+ =− xccλ ( Jx() c Jx () c) Jx()() c Rx c T ()λccJx() Jx () c( x+ −=− x c ) Jx ()() c Rx c TT λλccJx() Jx () c x+ =− cc Jx () Jx () cc x Jx ()() c Rx c

Ax22= b 2 T AJxJx2 = λcc() () c

xx1 = + TT bJxJxxJxRx1 =−λcc()() () ccc() () c

The Levenberg-Marquardt is another modification of the Gauss Newton method with an added regularization term, μcI. This added regularization terms helps to stabilize

T the solution of x+. With the addition of a constant term to the diagonal entries of J J, it expands the problems that can be solved to ones even with a rank deficient Jacobian

matrix. The local convergent properties of the Levenberg Marquardt are similar to the

Gauss Newton method. Below is the transformation of the Levenberg-Marquardt method into the common Ax=b form.

The rank of the Jacobian varied with depending on the time point within the action potential which the Jacobian was evaluated using the Gauss Newton and Damped

Gauss Newton methods. Therefore, all further conditioning and stability analysis was conducted with the Levenberg Marquardt method.

For conditioning analysis, the parameters theta and kappa were evaluated. Theta was defined as the arc cosine of the norm of the computed optical mapping signal divided by the optical mapping data. For this analysis, fabricated optical mapping data was used.

The condition number is defined as the ratio between the largest and smallest singular values of A. A is defined as the matrices A1, A2, and A3 derived in above. The condition number is a measure of how well conditioned a problem is to numerical computation. A problem with a low condition number is “well conditioned,” whereas a problem with a high condition number is “ill conditioned.”

Stability analysis included solving for x by 1) solving the normal equations, 2) using QR factorization of A by Householder, and 3) singular value decomposition (SVD) of A. Conditioning analysis of the matrix A3 was evaluated by varying the regularization parameter μs. A large regularization parameter is typically used when you do not have confidence the initial guess. Conditioning analysis of the problem included computing kappa and theta .Since the rank of the Jacobian varied with time, the condition number of the matrix the matrix A3, as computed by the Levenberg Marquardt method is presented

in Figure 2. 10. With this analysis, it was shown that the condition number varies with time (Figure 2. 10).

Figure 2. 10. Condition number as a function of time for the matrix A3 (Levenberg Marquardt method) To further analyze the condition number, the regularization parameter was varied between 1 and 300. A regularization parameter of 0 turns the Levenberg Marquardt method into the Gauss Newton. It was observed that the condition number decreases dramatically with the increase of the regularization parameter (Figure 2. 11).

Figure 2. 11. Condition number of matrix A3 as a function of regularization parameter (Levenberg Marquardt method)

Using the Levenberg Marquardt method, we were able to compute x+ using the normal equations, QR factorization and SVD. With a regularization parameter equal to one, the difference between the computed x+ by the normal equation and the SVD differed by the last three significant digits. However, with increasing regularization number, the difference between the normal equations and the SVD were only one or two significant digits.

Figure 2. 12. Comparison of computed and initial parameters

Gauss Newton and Damped Gauss Newton methods are not sufficient to solve the nonlinear optical mapping inverse model because the rank of the Jacobian changes with time. Regularization is necessary to reduce the condition number of the optical mapping inverse nonlinear least squares problem, and the Levenberg-Marquardt can be used. With a regularization parameter of at least one, the conductance parameters can be computed by solving the normal equations. The computed x+ is approximately the same for the range of regularization parameters analyzed (1 to 300) computed with the normal equations, QR factorization or SVD. In the future, OCT volumes obtained with an integrated Optical Mapping and OCT system can be used to further stabilize the optical mapping inverse model. The first step of the optical inverse model is the creation of the computational domain. The surface of sample can be generated from the OCT three dimensional dataset. The importance of the computational domain is to accurately represent boundary conditions in which to solve the diffusion equation.

2.4. Discussion

In summary, an automated algorithm was developed and validated for quantifying cardiac fiber orientation within OCT image sets. During this study, the sample surface was made relatively flat through the use of pins, and dehydration agents enhanced the visibility of fibers in OCT image sets. These are procedures that cannot translate directly to in-vivo imaging. Therefore, developing an OCT scanner with high axial and transverse resolution may alleviate the need for dehydration and fixation protocolsQuantifying fiber orientation in structurally intact excised preparations using OCT can potentially be used to correlate fiber orientation with electrical conduction properties (e.g. conduction velocity) and mechanical properties (e.g. strain analyses) of the myocardium in a variety of heart disease and arrhythmia models.

. Maintaining a relatively constant transverse resolution with depth will increase the visibility of fibers in depth. In addition, there is a need to extend this algorithm to quantify fiber orientation in three dimensions to accommodate samples with irregular fiber orientation patterns and samples with curved surfaces. The two dimensional fiber orientation algorithm can be extended to quantify fiber orientation within OCT volumes within an intact heart. Fiber orientation in three dimensions is described by two angles, an inclination angle, θ, and a transverse angle, φ.

The first step is to identify the surface from the three dimensional image set. To identify the surface automatically, the 3D image set is convolved with a Weiner filter for noise reduction and a threshold applied by analyzing the intensity histogram. Thereafter, the noise reduced 3D image set is convolved with 3D Sobel filters to compute directional

gradients. Gx, Gy, Gz are defined as the convolution of the x, y, and z volumetric Sobel filters with the 3D OCT image set. For each voxel, a second order tensor, Gxyz, is calculated. Following the principles of diffusion tensor MRI, the local fiber direction is the eigenvector, v, associated with the largest eigenvalue of Gxyz.

⎛⎞⎛v1coscosϕ θ ⎞ (, , )⎜⎟⎜ 2 cos sin ⎟ vi jk==⎜⎟⎜ v ϕ θ ⎟ ⎜⎟⎜ ⎟ ⎝⎠⎝v3sinϕ ⎠

θπ∈[]0, 2 φπ∈[0, 2 )

The dominant fiber direction will be computed within each subvolume by weighting a Kent probability distribution by the magnitude gradient, G(i,j,k) =

222 Gijkxyz(, , )++ Gijk (, , ) Gijk (, , ). Within a small local volume of the 3D data set, V, the dominant local direction of the fiber will be computed by taking the maximum of the

V angular distribution function, Fθ ,ϕ , a function of G(i,j,k), θ’(i,j,k), and φ(i,j,k).

⎡ 22⎤ expκβv123 (,,) i jk⋅+ x()() v (,,) i jk ⋅ x − v (,,) i jk ⋅ x V ( ⎣ ⎦) FGijkθϕ, = ∑ (, , ) (,ijkV , ) c

Fiber orientation is defined with respect to the surface. The sample surface will be downsampled by the same block size, V, used in the fiber orientation algorithm.

Normal vectors used to describe the down sampled surface will be used to carry out a global to local coordinate system transformation to assign final inclination and transverse angles to each subvolume. Implementation of this type of algorithm can be applied to

OCT three dimensional image sets of whole hearts (i.e. mouse hearts) or complex fiber organization (i.e. infarction border zones).

2.5. Conclusion

In this chapter we have demonstrated that OCT has the potential to visualize arrhythmogenic substrates, such as abnormal fiber orientation or myocardial infarctions.

Three dimensional image sets of optically cleared tissue were used to achieve these goals.

Applying optical clearing agents to increase fiber visibility ca not translate directly to in vivo application. However, OCT may provide additional structural information within small animal models of heart disease and arrhythmia mechanisms and therapy for correlation with electrical[31] and mechanical function.

Chapter 3: Characterization of ablation lesions using optical coherence tomography

3.1. Introduction

3.1.1. Radiofrequency Ablation Cardiac arrhythmias afflict millions of patients in the United States, resulting in frequent hospitalizations and high medical costs[1]. Radiofrequency ablation (RFA) has revolutionized the treatment of arrhythmias with over 80,000 procedures performed in the

United States each year[6]. The duration of RFA procedures may range from 2 to 8 hours, depending on the type of arrhythmia that is targeted. Currently ablation is monitored by indirect means by changes in catheter-tissue contact temperature, impedance, and intracardiac electrograms. A direct image of the cardiac tissue during

RFA could help guide the precise application of energy, monitor the successful formation of a lesion, visualize early effects of overtreatment, and potentially decrease procedure time. Reduction of procedure time would furthermore reduce radiation exposure to the patient and operator because the procedure is performed under fluoroscopy. In short,

RFA therapy could potentially benefit greatly from real time, direct imaging because it

could lead to decreased procedure time, fewer complications, and increased acute and chronic success rates.

3.1.2. Monitoring Radiofrequency Ablation Therapy

Currently, RFA procedures utilize fluoroscopy, static images from computed tomography merged with fluoroscopy, and/or noncontact, three-dimensional reconstructed mapping to visualize the position of catheters in the heart. Monitoring the delivery of RF energy to the endocardium is only by indirect means; that is, from change in the amplitude of intracardiac electrograms, change in tissue temperature, and change in impedance, all of which are measured at the tip of the catheter. These indirect methods of monitoring may result in delivering more ablation lesions than necessary to cure the arrhythmia and may also prolong procedural time. There are other approaches under investigation to monitor and guide RFA therapy, including magnetic resonance imaging

(MRI) [12, 13] and echocardiography [7, 16].

Radiofrequency ablation therapy is a standard of care for the treatment of specific arrhythmias, and this procedure would benefit significantly from a compact, high- resolution, real-time, image-based monitoring technology. We hypothesize that the emerging imaging technique of optical coherence tomography (OCT) has potential for monitoring cardiac radiofrequency ablation therapy and that OCT can validate that a lesion has been made, by distinguishing RF treated from untreated cardiac tissue.

OCT has been used for intra-coronary imaging[157], imaging of the myocardium within animal models of arrhythmias[31, 87, 89], and analyzing myocardial fiber structure using image analysis[31, 89]. In addition, OCT has been used to image laser aortic ablation, [158] laser ablation of brain, liver, kidney, lung and rectus abdominis

muscles [159], laser ablation of the esophagus[160] and high intensity focused ultrasound

(HIFU) ablation lesions in ventricular myocardium[161]. Previous studies have shown that the optical properties of heated myocardium (absorption, scattering and anisotropy coefficients) are significantly different from normal tissue [162-164] and that these optical properties can be extracted from OCT images [129] for the purpose of tissue classification. Birefringence is an optical property of highly organized tissue such as cartilage and muscle which results in modulation of the state of polarization of light propagating in the tissue. Polarization-sensitive OCT (PSOCT) has been used to measure tissue birefringence to assess tissue damage due to ablation therapy[56, 57]. Tissue birefringence can be observed as artifacts within conventional OCT images [165-167].

These artifacts have been used previously to assess collagen content within atherosclerotic plaques[166].

OCT can potentially address unmet clinical needs of cardiac radiofrequency ablation therapy by assessing the contact and contact angle of the RF catheter with the tissue, confirming that a lesion has been formed when RF energy is delivered, detecting early damage, and identifying structures for procedural guidance. Imaging to monitor tissue contact and contact angle can increase the efficiency of RF energy delivery. Acute success and efficacy of ablation lesions are determined through functional electrophysiology (EP) testing, to ensure that the lesion interrupts conduction. The ability to directly confirm that a lesion has been formed after energy delivery will eliminate ambiguity when EP testing shows that conduction interruption was not achieved, by eliminating the possibility the energy dose failed to result in a lesion. The

ability to detect early damage could enable titration of energy delivery and reduce complication rates.

Here, we report that ablation lesions in ventricular myocardium in vitro can be distinguished from untreated tissue using OCT imaging, by detecting changes in tissue birefringence, optical properties, and tissue architecture. We also observe that overtreated lesions exhibit additional characteristic features. This work sets the foundation for investigating the feasibility of real-time monitoring of RFA therapy using OCT.

While the axial field of view of OCT imaging is too small to encompass an entire RFA lesion, real time monitoring of RFA therapy with high-resolution imaging could provide a direct visual the heart surface, visual and confirmation of ablation lesion formation and feedback to titrate RFA dosage to avoid complications and minimize procedure time.

3.2. Methods

3.2.1. Feasibility Testing

Preliminary studies were conducted to evaluate if OCT can distinguish necrotic tissue from untreated tissue. Freshly excised ventricular wedges from swine hearts were used for in vitro characterization of ablation lesions within OCT images. Following onset of general anesthesia a lateral thoracotomy was performed and the heart was rapidly excised and placed in ice-cold phosphate buffered saline (PBS). Individual sections of cardiac tissue were placed in a bath with PBS maintained at 37oC with super-perfusion flow. All animal studies were conducted according to protocols approved by the

Institutional Animal Care and Use Committee of Case Western Reserve University.

Ablation lesions were created with a constant power protocol (15-45W) with a maximum temperature of 95oC for 60 seconds to generate a range of lesion depths. RFA

lesions were created with a 7Fr RFA catheter and EPT-100XP Generator (Boston

Scientific). Maximum and average temperature, impedance, and power were recorded for each experiment. After administering RFA, a pair of pins was placed on either size of the ablation lesion.

OCT imaging was conducted on a system with a 1310nm TDOCT system with 10 micrometer axial resolution. A long working distance (5.2cm) scanner was designed using Zemax, optical design software. The optical design and system point spread function (PSF) are shown in Figure 3. 1. The scanner maintains an 18 micrometer, full width at half maximum (FWHM) spot size over an 8 lateral scan range. Eight millimeter

OCT scans were taken to encompass the lesion and neighboring viable tissue.

Figure 3. 1. Long working distance OCT scanner. A) Zemax optical design of scanner. B) Point spread function.

Staining with triphenyltetrazolium chloride (TTC) was used to identify necrosis and quantify lesion size. TTC is a vital staining for assessing dehydrogenase enzyme activity and has been widely used for assessing acute injury to differentiate necrosis from viable tissue. Previous studies have shown that an animal has to survive 24 hours post injury, preferably 3-4 days, for histological signatures to distinguish necrosis from viable myocardium[168]. In addition, histological analysis can be time consuming, expensive, and requires an expert for analysis. Tetrazolium salts produce colored precipitates in the presence of an intact dehydrogenase system. Necrotic tissue lack dehydregease active and fails to stain. This fast and cheap method allows macroscopic evaluation of myocardial tissue to distinguish necrotic tissue from viable tissue within acute injury.

Within the preliminary experiments, the tissue was sliced in the direction of the OCT B- scan, as indicated by the pins and incubated in 1.0% TTC in PBS for 15 minutes at room temperature. The maximum length and width were recorded for each lesion.

Figure 3. 2. a) Gross image of Endocardium of swine right ventricle with RFA lesions. Ablation lesions created with a constant power protocol. b, c, e, g) representative OCT images of ablation lesions. d) TTC vital stain of ablation lesion shown in e). f) TTC vital stain of ablation lesion shown in (g). TTC stains necrotic tissue white and viable tissue red. Image analysis tools were developed to extract parameters quantifying tissue scattering properties, tissue heterogeneity, and presence of tissue birefringence for the purpose of tissue classification. A preprocessing step was conducted for each image to reduce noise by convolving with a Wiener filter and correct for the focusing effects of the objective lens. Thereafter, each image was decimated to 190 x 24 pixels. The region of interest (ROI) analyzed was 600 microns in length in the axial dimension, starting 200

µm below the sample surface to avoid surface reflections and the endocardial layer.

Iz()=− Ihzot ()exp2( μ z) −1 ⎛⎞2 ⎛⎞zz− cf hz()=+⎜⎟⎜⎟1 ⎜⎟z ⎝⎠⎝⎠R where

Io :i Backscattered _ ntensity

μt : attenuation_ coefficient z: axial_ depth

zcf : axial_ location__ of focus

zR : Rayleigh rang e

The model takes into account the focusing effect of the objective lens, and the coherence length of the light source. The rate at which the intensity falls is attributed to the attenuation coefficient, µt, which is equal to the sum of the scattering and absorption coefficients. By computing the logarithm of the single scattering model, we obtain a

linear function of the attenuation coefficient yz()= −+ 2μt zn( ) c . The axial distance, z, is a function of index of refraction, n. A constant index of refraction of ventricular muscle, 1.382[170], was assumed for all tissue samples in this study. The birefringent myocardial tissue produces a polarization artifact within OCT images; and this artifact confounds the measurement of the attenuation coefficient. Therefore, the slope was

' referred to as the signal decay rate, μt . Columns in the decimated image, representing 20 averaged A-scans of the original image, were fit to the linear model. Using linear least

' square fitting, estimates of the signal decay rate, μt , and intercept, c, were obtained. To assess tissue heterogeneity, the correlation coefficient (the R value of the linear regression analysis) was computed as a measure of how well the averaged axial scans fit the linear model. The three parameters (signal decay rate, intercept, and correlation) gave an indication of tissue scattering, reflectivity, and heterogeneity respectively. Analysis of variance was conducted to evaluate if OCT derived parameters can be used to distinguish viable tissue from necrotic tissue. A p value less than 0.05 was considered statistically significant. Results are presented as mean (1 standard deviation).

Figure 3. 3: Classification of lesions and untreated tissue. Using signal decay rate and correlation coefficient, separation between ablation lesions and viable.

Ablation lesions created with a constant power protocol produced a large range of lesions sizes, with an average lesion depth of 3.45(1.87) mm. Electrode temperature reached an average of 75.8 (10.1)oC. Signal decay rate decreases with increasing lesion depth (Figure 3. 5), and has a nearly linear relationship (R2=0.746). Intercept was not statistically different between lesions and untreated sites (p=0.67).

Within OCT imaging of endocardial ablation lesions on a New Zealand White rabbit, OCT was able to visualize adverse effects due to overtreatment. OCT images were acquired after the occurrence of a steam pop. Craters and cardiac perforation is observed within OCT images and coagulum visible on the RFA catheter(Figure 3. 6).

Using the mean and standard deviations of the image analysis parameters derived within the preliminary studies, a power analysis as conducted, using the PS Power and

Sample Size Calculator software developed at Vanderbilt. With a ratio of one third control samples to lesions, 39 images are required to have a study with 99% power and

0.01 type I error probability. In addition, the study protocol was changed from power controlled to temperature controlled to mimic clinically relevant parameters and decrease the number of steam pops.

3.2.2. In vitro ablation protocol

Freshly excised ventricular wedges from swine hearts were used for in vitro characterization of ablation lesions with OCT imaging. Following the onset of general anesthesia, a lateral thoracotomy was performed, and the heart was rapidly excised and placed in ice-cold phosphate buffered saline (PBS). The right ventricular free wall, left ventricular free wall, and septum were dissected and placed in ice-cold PBS, up to 2 hours, until the start of RFA. All animal studies were conducted according to protocols approved by the Institutional Animal Care and Use Committee of Case Western Reserve

University.

Dissected ventricular wedges were placed in a bath with PBS maintained at 37oC with super-perfusion flow (Figure 3. 7-A). During the application of RF energy, all samples were submerged in 3 cm of PBS so that the RF catheter tip is completely submerged and operates normally. A series of ablative lesions were created with a temperature controlled (70oC) protocol with a maximum delivered power of 50W and maximal duration of 60 seconds, using the Maestro 3000 RFA generator (Boston

Scientific). Endocardial lesions were created using a 7Fr, 4mm tip Blazer II catheter

(Boston Scientific), applying RF energy for 10 seconds (n=19), 20 seconds (n=22), 30 seconds (n=22), and 60 (n=29) seconds distributed over 7 hearts for a total of 92 lesions.

An additional set of endocardial ablation lesions were created on ventricular tissue with the same temperature controlled protocol. After creating ablation lesions, these samples were subsequently volumetrically imaged by OCT for assessment of myofiber visibility within OCT volumes[31, 89] and identification of gaps within linear ablation lines.

Figure 3. 7. Experimental setup for in vitro characterization of ablation lesions using OCT. A) In vitro radiofrequency ablation lesions created on excised ventricular wedges in a temperature controlled bath with super-perfusion flow. B) Gross pathology of ablative lesions on the endocardial surface of swine right ventricle. (Pins demarcate ends of ablative lesions) C) Optical coherence tomography (OCT) system. D) Representative OCT image of a seven millimeter image of ablative lesion and adjacent tissue E) TTC stains of ablative lesion shown in panel D)White necrotic tissue of ablative lesion and red viable tissue demonstrated 3.2.3. Optical Coherence Tomography Imaging

After administering RFA, a pair of pins was placed along the maximal dimension of each ablation lesion (Figure 3. 7-B). Imaging was conducted with a time-domain OCT system described previously[171] with a light source centered at1310 nm with a 70 nm bandwidth (Figure 3. 7-C). The axial and lateral resolution of the system was approximately 10 and 18 micrometers respectively (in air). Seven-millimeter OCT scans were recorded to encompass the lesion, and seven-millimeter control images were recorded of areas where no RFA energy was delivered (480 A-lines per image). These data were used for the lesion detection study.

Three dimensional image sets of ventricular wedges with RFA lesions were acquired with a microscope-integrated 1310nm Fourier Domain OCT (FDOCT) system.

The axial and lateral resolution of this system was approximately 10 micrometers (in air)[141]. Spectral interferograms were acquired with a linear in wavenumber (k=2π/λ) spectrometer[50] onto a 1024 pixel line scan camera (Goodrich) spectrometer, acquired at

40 images per second (1000 A-lines per image). The system has a 2mm -6dB imaging range and 110dB signal-to-noise ratio. Three-dimensional image sets were 4 x 4 x 4.3 mm3 in dimension, with 400 images per volume. These data were used to investigate the appearance of gaps

3.2.4. Validation

Staining with triphenyltetrazolium chloride (TTC) was used to validate lesion formation and to quantify lesion size[172]. The tissue was sliced in the direction of the OCT B- scan, as indicated by the pins and incubated in 1.0% TTC in PBS for 20 minutes at room temperature. The TTC stained samples were digitized with a calibration marker. Using the software package Image J (NIH), lesion depth, width, and area were recorded for each lesion.

3.2.5. Image Analysis

Two main characteristics of OCT images of RFA lesions, compared to untreated tissue, were observed and targeted for quantitative analysis. First, the intensity increases due to increased scattering. Second, the birefringence artifact is eliminated due to decreased tissue birefringence. Image analysis tools were developed (using Matlab 7.4,

Mathworks Inc.) to quantify these two parameters for the purpose of distinguishing ablation lesions from untreated tissue. Preprocessing was conducted for each image to

reduce noise by convolving with a 5 x 5 Wiener filter and removing the noise floor from

the image. The images were flattened by detecting the tissue surface with using a

threshold. Thereafter, each image was decimated to 190 x 24 pixels (Figure 3. 8-a-c) to reduce computation and speckle noise in axial scans (Figure 3. 8-g-i). The region of interest (ROI) analyzed was 525 microns in length in the axial dimension, starting 100

µm below the sample surface to avoid surface reflections and the endocardial layer. The lateral region of interest was the entire 7mm image for untreated tissue, and the area in between the pins for treated tissue. The intensity parameter was calculated as the mean intensity of the decimated and flattened OCT image within the ROI.

A dark band was consistently observed within images of untreated tissue (e.g.

Figure 3. 8-a,b). To demonstrate that the observed bands were birefringence artifacts and not tissue structures, a sample site was imaged while the polarization state of the sample arm light was changed using a polarization controller. Figure 3. 8 shows two images of the identical site illuminated by different polarization states. It can be clearly observed that the band changes location in a polarization-dependent way.

In order to automatically detect the birefringence artifact, a Laplacian of Gaussian

(LoG) was implemented. The LoG filter has been previously used to identify regions that are brighter or darker than their surroundings[173]. The LoG is a linear filter and is a combination of a Laplacian operation and a Gaussian filter. The Laplacian estimates the second derivative of the signal. The Gaussian filter is used to smooth the data and reduce the contribution of noise to the second derivative. The parameters of the LoG are the kernel dimensions and the standard deviation for the Gaussian. LoG with a 20x1 pixel kernel size and 0.5 pixel standard deviation was convolved with the flattened and decimated OCT image. The gradient strength parameter, representing the presence of the birefringence artifact, was defined as the mean pixel value within the ROI of the LoG- filtered image.

The single scattering model for OCT images was used to analyze tissue scattering and tissue heterogeneity. Within the first few hundred microns below the tissue surface, the intensity distribution in the axial direction of an OCT image can be approximated by a first order model of scattering, where the signal power decreases in an exponential manner[129, 169]. The model takes into account the focusing effect of the objective lens, and the coherence length of the light source. The rate at which the intensity falls is

attributed to the attenuation coefficient, µt, which is equal to the sum of the scattering and absorption coefficients. By computing the logarithm of the single scattering model, we

obtain a linear function of the attenuation coefficient yz()= −+ 2μt zn() c . The axial distance, z, is a function of index of refraction, n. A constant index of refraction of ventricular muscle, 1.382[170], was assumed for all tissue samples in this study. The birefringent myocardial tissue produces a polarization artifact within OCT images; and this artifact confounds the measurement of the attenuation coefficient. Therefore, the

' slope will be referred to as the signal decay rate, μt . Columns in the decimated image, representing 20 averaged A-scans of the original image, were fit to the linear model.

' Using linear least square fitting, estimates of the signal attenuation rate, μt , were obtained. To assess tissue heterogeneity, the correlation coefficient (the R value of the linear regression analysis) was computed as a measure of how well the averaged axial scans fit the linear model. The lateral variability of the correlation coefficient was defined as the standard deviation (SD) of the R values obtained for each column of an image. The mean intensity was the average pixel value within the ROI. The imaging depth was defined as the depth to which the intensity falls to 1/e of the intensity of the starting position within the ROI.

To assess tissue heterogeneity, average axial scans were fit to a line and the correlation coefficient (the R value) of the linear regression analysis was computed. To identify holes within the image using morphological processing. The matlab function,

IMFILL, was used where holes are defined as an area of dark pixels surrounded by light pixels. The intensity difference was computed between the filled image and original image.

3.2.6. Statistical Analysis

Analysis of variance (ANOVA) and receiver operator characteristic curves (ROC) were used to determine whether the quantified tissue classification parameters can be used for binary differentiation of untreated tissue from RFA lesions. Results are reported as mean

(95% confidence interval). The software package Origin 8.0 was used to conduct statistical analysis and a p-value less than 0.05 was considered statistically significant.

3.3. Results

A total of 92 images of lesions and 30 control images were distributed between the septum, right ventricle, and left ventricles of 7 swine hearts. These images were analyzed with OCT imaging Figure 3. 7-C, D) and subsequently with TTC staining (Figure 3. 7-E).

As shown in Figure 3. 10, lesions created in this in vitro model follow the predicted trend where temperature increased over time, approaching the target temperature by 60 seconds. Impedance decreased over time (Figure 3. 10-B), and power remained roughly constant (Figure 3. 10-C). This resulted in lesions depths, as determined by TTC, which increased as a function of RF energy delivery time (Figure 3. 10D). Thus, the created lesions accurately modeled clinical RFA lesions.

3.3.1. Tissue Classification

OCT images of untreated sites had a characteristic polarization-dependent band, due to the bifringence property of highly organized myocardial tissue. Within volumetric image sets of linear ablation lines, a gap of viable tissue between two ablation lesions was identifiable within OCT images as a birefringent dependent band (Figure 3. 11).

By visualizing the plane parallel to the surface of three dimensional image sets of samples fixed in formalin, fiber organization was observed within untreated sites, and fiber organization was not visible within the RFA lesion (Figure 3. 12-C). Areas with

RFA lesions lost visible fiber structure within OCT volumes and also lost the banding appearance within B-scan images (Figure 3. 12-B).

Figure 3. 12 Loss of fiber organization within lesion. A) Microscope image of representative lesion and adjacent untreated tissue, formalin fixed. Black box denotes OCT field of view 4x4 mm2. Yellow dotted line indicates lesion boundary. Black dotted line indicates location of B-scan shown in panel B. B) OCT B-scan demonstrates absence of birefringence band in necrotic tissue of RF lesion. Birefringence band is present on the left, within the area of untreated tissue; whereas, dark band is absent within area of lesion. Birefringence band is also demonstrated in double hump within average axial scan (red) in the area with no lesion. C) Slice parallel to the sample 480µm below sample surface, showing fiber organization within area without a lesion and lack of fiber organization within the lesion. Images acquired with microscope integrated Fourier Domain (FDOCT) system.

RFA lesions (Figure 3. 14) were characterized by increased intensity and an absence of a birefringence dependent polarization artifact band.

100

To evaluate the ability of each image analysis parameter to distinguish lesions from untreated tissue, a receiver operator characteristic analysis was conducted. The mean value of analysis parameters gradient strength and mean intensity for each site was used for analysis. The lateral region of interest for lesions was defined as the area between the pins.

101

All Samples mean (95% CI) p-value AUC untreated lesion attenuation 52.5 (2.0) 37.4 (1.4) 5.71E-08 0.85 R -0.9 (0.01) -0.76 (0.02) 8.70E-08 0.73 R SD 0.074 (0.012) 0.232 (0.019) 6.97E-11 0.80 Intensity 35.8(1.7) 43.5(1.0) 2.82E-04 0.73 LoG 2.0(0.25) -1.2(0.1) 5.93E-15 0.94 imaging depth 0.194 (0.013) 0.269 (0.006) 6.64E-06 0.79

Table 3. 1. Binary discrimination of lesions and untreated tissue To evaluate the ability of each image analysis parameter to distinguish lesions from untreated tissue, a receiver operator characteristic analysis was conducted. The mean value of analysis parameters gradient strength and mean intensity for each site was used for analysis. The lateral region of interest for lesions was defined as the area between the pins. The area under the curve was largest for the gradient strength, 0.94.

The gradient strength was significantly different between untreated samples 2.0 (0.25) and ablation lesions -1.2 (0.1), p=5.93x10-15. By choosing a threshold that produces the maximum accuracy, 0.78, the gradient strength had a high sensitivity and specificity,

94.5% and 86.7% respectively. When analyzing the data according to tissue type, the gradient strength had an AUC of 0.99, 0.95, and 0.97 for the left ventricle, right ventricle, and ventricular septum and a sensitivity and specificity for the right ventricle (94.6%,

93.3%), left ventricle (96.4%, 100%) and ventricular septum (96.3%, 100%). Mean intensity was significantly increased, p=2.82x10-4, between untreated samples 35.8 (1.7) and ablation lesions 43.5 (1.0). However, mean intensity was not a strong classifying parameter with an AUC of 0.72 and a threshold of 28.7 that produces maximum accuracy resulted in a sensitivity of 94.6% and specificity of26.7%. This is demonstrated Figure 3.

15, where the main parameter separating untreated sites from ablation lesions was the gradient strength.

102

103

A mean 24W of RF energy was delivered to the wedges, resulting in an average lesion depth of 1.5 (0.10) mm and surface width of 4.5 (0.14) mm, determined by TTC staining. The surface width indicated by the pins was 3.9 (0.08) mm. The width of ablation lesions confirmed by TTC vital staining was slightly larger (0.67 mm) than the distance between pins used to identify RFA lesions within OCT images. Thus, the area used for analysis of RFA lesions within OCT images was confirmed to be necrotic tissue.

Image analysis parameters gradient strength and mean intensity did not show a relationship with lesion depth.

3.3.2. Visualization of Overtreatment

Within a subset of OCT images of RFA lesions (n=34), disruptions were visible within the myocardium (Figure 3. 16). These disruptions were attributed to overtreatment; however none of the procedures produced craters, tearing of the endocardial surface, or audible steam pops, indicative of endocardial or myocardial rupture. The standard measurements of mean temperature (p=0.997), impedance (p=0.467), and power

(p=0.488), were not significantly different between lesions without visible disruptions and over-treated lesions. Image analysis parameters, correlation coefficient and filling intensity were statistically different (p<0.001) between adequately treated lesions and over treated lesions.

104

Four ablation lesions and seven control sites were imaged in the left atrial appendage. Representative images are shown in Figure 3. 18. The contrast between the endocardial layer and the myocardium is decreased after energy delivery. Unlike the untreated myocardium in the ventricle, shown in Figure 3. 13, the untreated atria is not characterized by a polarization artifact band.

105

Radiofrequency ablation (RFA) causes thermal damage due to resistive heating, producing an area of coagulation necrosis[6]. Thermal damage of the myocardium has been shown to cause changes in the optical properties of tissue, in particular anisotropy coefficient, scattering coefficient[162-164], and birefringence[174]. Our results show that using a conventional, single detector, OCT system there are measurable tissue characteristics that are markers of RFA lesion formation. This demonstrated the potential for the use of OCT to address an important unmet clinical need. Image parameters

106

related to changes in optical properties and tissue architecture can be used for binary discrimination of RFA lesions from the untreated myocardium. OCT image-derived parameters were used to differentiate untreated sites from RFA lesions.

Viable myocardial tissue is birefringent, which resulted in a polarization- dependent banding artifact within OCT images. Tissue birefringence was lost within ablation lesions. This result is consistent previous observations using polarization- sensitive OCT (PSOCT) of thermally damaged skin and tendon where tissue birefringence was lost after heating [56]. The birefringence artifact visible in OCT images of untreated myocardium provided an effective contrast mechanism between untreated tissue and non-birefringent ablation lesions. Therefore, with a conventional

OCT system, accurate binary tissue classification was possible without explicitly quantifying tissue birefringence by use of the gradient strength parameter. Although the birefringence band’s location can vary between control samples, the gradient strength quantified the presence of the birefringence band and had a small variability within control samples. It is expected that use of PSOCT will improve detection of tissue birefringence, and also provide artifact-free images that will allow for additional analysis such as changes in tissue attenuation. Therefore, in the future, we will evaluate the use of

PSOCT to measure tissue scattering and birefringence to increase the accuracy of tissue classification.

Previous work has shown that scattering increases within ablation lesions[162].

Because signal intensity is related to the scattering coefficient, we expected a higher mean signal intensity for ablated tissue compared to untreated tissue. Signal intensity increased significantly within all tissue types after RFA (p=2.82x10-4). The absence of

107

the birefringence dependent band within treated tissue may contribute to the significant increase in image intensity. Therefore, increased signal intensity with the application of

RF energy may be contributed to both an increase in scattering and an absence of the birefringence dependent band. However, there may be intrinsic differences in optical properties between untreated tissue and ablation lesions that warrant further investigation.

The mean intensity within ventricular septum samples were significantly lower (p=0.002) than right and left ventricular samples. Lesions created on the ventricular septum were conducted after experiments with the right and left ventricles were completed, but within the 2 hour period. Therefore, the ventricular septum was left in PBS longer before the start of RFA compared to right and left ventricles. This prolonged time within PBS may have contributed to the decreased intensity of the ventricular septum samples

OCT image-derived parameters based on the single-scattering model were used to differentiate untreated sites from RFA lesions. These parameters are related to the underlying optical properties of the tissue and tissue architecture. For our analysis, a single-scattering model was used which neglects polarization effects, and which assumes a homogenous sample. Viable myocardial tissue is birefringent, which resulted in a polarization-dependent banding artifact within OCT images. Tissue birefringence was lost within ablation lesions. This result is consistent previous observations using polarization-sensitive OCT (PSOCT) of thermally damaged skin and tendon where tissue birefringence was lost after heating. [56]. The birefringence artifact visible in OCT images of untreated myocardium provided an effective contrast mechanism between untreated tissue and non-birefringent ablation lesions. Therefore, with a conventional

OCT system, accurate binary tissue classification was possible without explicitly

108

quantifying tissue birefringence by use of the gradient strength parameter. However, the presence of a polarization artifact confounded the measurement of the attenuation coefficient in untreated tissue, which resulted in additional source of signal attenuation not accounted for in the model. The location of the birefringence band is dependent on multiple factors including the orientation of the beam and the orientation of the muscle fibers. Therefore, the signal decay rate and imaging depth parameters depend partially on the presence and the location of the birefringence artifact, likely decreasing their value as complementary classifiers to the gradient strength.

Disruptions within the myocardium were observed in a subset of OCT images, which we believe to be precursors to steam pops. It is current practice to use the increase of tissue-electrode impedance as a signal to detect overtreatment of the tissue and coagulum development[6]. However, previous studies have shown that adverse effects such as steam pops are not always associated with a large change in impedance[175].

More than ever, this is important for saline irrigated RFA catheters. Intracardiac echocardiography has also been used to monitor ablation therapy, which may allow titration of RF energy to reduce incidence of embolic events due to over-treatment. This technique provides real-time monitoring, but relies on the visualization of microbubbles[7], an indirect measure of tissue state, to guide the treatment protocol. As demonstrated in Figure 8, high resolution, subsurface imaging using OCT can provide direct visualization of early damage to the myocardium, which may be precursors to steam pops and crater development. This can potentially be a valuable tool to titrate RF dosage and reduce complications.

109

This study demonstrates the potential for OCT to be used as a monitoring tool for cardiac RFA therapy. Real time imaging modalities have potential clinical utility in monitoring and guiding RFA procedures by providing a visual of the heart-catheter interface, visualization of normal areas of the heart to avoid during ablation, validation that a lesion has been made, and detecting precursors to complications. Our results do not provide evidence that OCT may be suitable for assessing depth of cardiac RFA lesions. We demonstrated that OCT can be used to differentiate ablation lesions from untreated tissue and visualize potential complications due over treatment. These results motivate the next step toward translation of this technology. Future in vivo investigations will require catheter based imaging using PSOCT for assessing dynamics due to RF energy delivery. A contact OCT catheter will displace blood from the OCT field of view, allowing the OCT probe beam direct access to the tissue surface. Saline irrigated RFA catheters are becoming standard during complex RFA procedures and could be used in combination with contact OCT catheters to further displace blood. A forward-imaging probe in contact with the tissue will largely mitigate motion artifacts during in vivo imaging. High speed imaging will further mitigate motion artifact as well as any artifacts due to changes in myocardial optical properties as a function of time during the cardiac cycle. This will enable future studies to evaluate the ability for OCT to discriminate viable tissue from ablation lesions in the presence of blood, with varying catheter contact angle, and in vivo. Catheter-based imaging will allow for real-time monitoring during

RF energy delivery, assessment of tissue contact, and observation of the time dependence of image analysis parameters to evaluate whether OCT guidance can identify early markers of complications and observe lesion formation in real time.

110

3.5. Conclusion

This study demonstrates that optical coherence tomography has the potential to monitor the formation of radiofrequency lesions in the heart. Our experimental results validate that using OCT imaging, RFA lesions can be distinguished from the untreated endocardium. Changes in optical properties and loss of birefringence provide intrinsic contrast between untreated tissue and ablation lesions. In addition, potential predictors of complications such as crater formation are visible within OCT images.

A direct image by OCT has the potential to guide the precise application of energy, avoid normal cardiac structures where ablation could be harmful, ensure adequate tissue contact during energy delivery and provide real-time formation of successful lesions. Importantly, this may decrease the procedure time and radiation exposure to the patient and physician. Furthermore, real-time feedback from OCT during RFA therapy may enable decreased complication rates. Therefore, these data set the foundation for

OCT as an imaging modality for future real-time direct monitoring of cardiac RFA therapy.

111

Chapter 4: Toward Guidance of Epicardial Radiofrequency Ablation Therapy using Optical Coherence Tomography

4.1. Introduction

Termination of an arrhythmia by RFA is typically targeted from the endocardium.

However, it is estimated that 10–20% of arrhythmia circuits are in the deep myocardium or epicardium[176]. These rates are higher for ventricular tachycardia (VT). Within a study of 257 consecutive patients, twenty eight, thirty seven and twenty four percent of patients with post myocardial infarction, Chagas disease, and idiopathic dialated cardiomyopathy, respectively, had epicardial substrates for VT [177].

4.1.1. Epicardial Ablation

Current treatment methods for ventricular tachycardia include antiarrhythmic drugs, implantable defibrillators and catheter ablation[178]. Implantable defibrillators and antiarrhytmic drugs palliate arrhythmia symptoms, whereas ablation therapy is a

112

curative method. Because of the large percentage of epicardial circuits for VT, percutaneous epicardial ablation is an attractive method. Acute success of epicardial ablation for VT is between 60 and 75% using standard RF catheters[179]. In a study of

112 patients with Chagas’ VT, the 3-year recurrence rate of patients that received simultaneous epicardial ablation and endocardial ablation was 30%[180]. One factor limiting success rates is attributed to RF energy delivery in the presence of epicardial fat.

Fat attenuates RF energy, resulting in limited lesion depth[177] and may mimic infarction during electro-anatomical mapping[181]. In addition, complications can arise, requiring early termination of the procedure. A major complication that has been encountered in epicardial RFA is vascular injury and thrombosis due to energy delivery in close proximity to a coronary vessel. A previous study by d’Avila et al showed that the susceptibility of damage of coronary vessels is inversely proportional to vessel diameter.

Where it is hypothesized that larger vessels are protected due to increased blood flow[182]. For coronaries within 1mm of an ablation lesion, small vessels showed severe injury, the presence of thrombosis and intimal hyperplasia[183]. One of the most important criteria for a successful RFA lesion is good and continuous catheter-tissue contact. However, this is difficult to assess during RF delivery.

4.1.2. Pre-procedural imaging

Currently, procedures are guided by real time fluoroscopy, electrograms, electrode temperature, and tissue-electrode impedance. However, a real-time view of the epicardial surface is not available for procedural guidance and identification critical structures. The use of pre-procedural imaging has been demonstrated to be valuable for planning complex RFA procedures, obtaining a detailed anatomy of the heart, identifying

113

areas of infarction using MRI[184, 185] and echocardiography[20], and epicardial fat using computed tomography[14, 15]. Three dimensional volumes of the heart structure and segmented volumes of fat and infarction can be integrated with 3D voltage mapping for improved procedural planning. These volumes are obtained hours to a day before the procedure, are low resolution and are static. This results in displacement errors once correlated with 3D voltage maps[178] or real time fluoroscopy, limiting the utility of merged images. Directly before the start of ablation therapy, coronary angiography may be conducted to provide a visual of coronary vessels. If the site of arrhythmia focus is near a coronary, a risk stratification analysis is conducted before the start of RF energy delivery. There is a clinical need to assess epicardial substrates with high resolution and in real-time, to reduce procedural complications by avoiding coronary vessels and increasing efficacy, by titrating RF energy dosage in the presence of epicardial fat, validating lesion formation and assessing RF catheter-tissue contact.

4.1.3. Optical Coherence Tomography

OCT can address unmet clinical needs of epicardial radiofrequency ablation therapy by assessing the contact and contact angle of the RF catheter with the tissue, confirming that a lesion has been formed after energy delivery, detecting early damage and identifying structures for procedural guidance. We have previously demonstrated the ability to monitor thermal ablation of the myocardium by radiofrequency ablation (RFA) using OCT[186] (Chapter 3). Viable tissue is characterized by a polarization artifact dark band within conventional OCT images due to the birefringence property of the highly organized myocardial tissue. Previous studies have also been conducted to evaluate laser ablation of porcine myocardium [57] and thermal heating of porcine skin[56] with

114

polarization sensitive OCT (PSOCT), observing changes in birefringence properties.

OCT has been demonstrated to visualize critical structures related to electrical conduction, including the purkinje network[87], and imaging the fast and slow pathways in the atrial-ventricular (AV) node[31, 88], myofiber organization[31, 89] in animal models and in vitro preparations of human tissue[90]. In addition, OCT has been used extensively for intravascular applications [86, 187], analyzing blood flow and detecting small blood vessels in ophthalmology[188], and characterizing adipose tissue in breast cancer[189, 190].

We hypothesize that OCT can be used for monitoring and guidance of epicardial ablation procedures. Providing a cross sectional view of coronary vessels will allow evaluation of vessel lumen size. With this information, the clinician will have more information to assess if energy can be delivered safely. Assessing tissue contact, validating lesion formation, and visualization of epicardial fat will reduce ambiguity in interpreting electrograms and enable titration of RF energy dosage to ensure effective energy delivery. Here we report the results of OCT imaging of normal untreated myocardium, ablation lesions, epicardial fat, and coronary vessels, and describe the distinguishing characteristics, including differences in optical properties and tissue architecture, and assessment of tissue contact with catheter based imaging[191]. This sets the foundation for developing OCT as a guidance tool for epicardial radiofrequency ablation. Real time imaging of epicardial RFA therapy with OCT can potentially provide a direct visual of the epicardial surface, assess contact, provide visual confirmation of energy delivery and feedback to reduce complications and titrate dosage by visualizing critical structures of interest.

115

4.2. Methods

4.2.1. Sample Preparation

Freshly excised ventricular wedges from swine hearts were used for characterization of epicardial structures and ablation lesions with OCT imaging.

Following the onset of general anesthesia, a lateral thoracotomy was performed, and the heart was rapidly excised and placed in ice-cold phosphate buffered saline (PBS). Eight hearts were used for this study. The right ventricular free wall and left ventricular free wall were dissected and placed in ice-cold PBS, up to half an hour, until the start of imaging. All animal studies were conducted according to protocols approved by the

Institutional Animal Care and Use Committee of Case Western Reserve University.

For characterization of epicardial RF lesions, dissected ventricular wedges were placed in a bath with PBS maintained at 37oC with super-perfusion flow (Figure 4. 1-A).

A series of ablative lesions were created with a temperature controlled (60oC) protocol with a maximum delivered power of 50W and duration of 60 seconds, using the Maestro

3000 RFA generator (Boston Scientific). Epicardial lesions were created using a 7Fr,

5mm tip Blazer II catheter (Boston Scientific).

116

4.2.2. Imaging Protocol

Three dimensional image sets of swine ventricular epicardium were acquired with a microscope integrated spectral domain OCT (SDOCT) system with a light source centered at 1310nm with 70nm bandwidth (InPhenix). Spectral interferograms were acquired with a linear-in-wave number (k=2π/λ) spectrometer[50] onto a 1024 pixel line scan camera (Goodrich) spectrometer, acquired at a 40kHz line scan rate. The system has a 4.3 mm imaging range, 2mm -6dB fall off range, and 110dB sensitivity. The measured

axial and lateral resolution of the system was 16 and 12 micrometers (in air)

respectively[141]. Each image was 4mm in transverse length, 1000 lines per image, and

512 pixels per line. Each volume consisted of 400 images. Assuming an index of

refraction of 1.38 for ventricular tissue[170] the dimensions the volumes were 4mm x

4mm x 3.11mm (L, W, H), with a corresponding pixel resolution of 4µm, 10 µm, and 6

µm respectively. Summed voxel projection was used for rapid visualization of the three

dimensional image sets and planes parallel to the sample surface were obtained by

117

detecting the surface with an intensity threshold and digitally flattening the tissue surface[31].

Imaging of epicardial substrates with a flexible catheter probe was conducted to show the feasibility of OCT imaging of epicardial substrates using a probe suitable for in vivo imaging. The catheter probe has been described in Chapter 3 [192]. Briefly, the probe is a contact, forward imaging probe that provides cone scanning. The probe uses a grin lens for focusing, a Risely prism to deflect the beam off axis, and applying torque on the fiber rotates the entire optical assembly to provide cone scanning (Figure 4. 2). An optical window isolates the probe optics from the environment. The probe sheath and end cap are polymers which are biocompatible and can operate in temperatures experienced in RF therapy, up to 95oC. The rigid end length is 18mm and the outer diameter is 2.5mm. The scan diameter is 2mm, which results in a 6.28mm lateral scanning range. The probe maintains a FWHM spot size of less than 30µm over an entire

1mm working range from the optical window. Images with the forward imaging catheter were acquired at 40 kHz line scan rate, 2000 lines per image, and 512 pixels per line.

This corresponds to a 6µm and 3.1 µm pixel size in the axial and lateral dimensions respectively. A correlation based method [193] was used to correct for non-uniform scanning rates, removing highly correlated axial scans.

118

4.2.3. Validation

For all volumetric images, a microscopic image of the surface of the sample was acquired and registered with the OCT volume. Staining with triphenyltetrazolium chloride (TTC) was used to identify necrotic and viable tissue for samples with ablation lesions. TTC is a vital staining where viable tissue stains red and necrotic tissue stains white. TTC vital staining is a standard procedure for assessing acute necrosis.

Immediately after the imaging protocol was completed, the tissue was sliced through the center of the ablation lesion and incubated in 1.0% TTC in PBS for 30 minutes at 37oC.

4.3. Results

Volumetric OCT images of epicardial structures on the right and left ventricles with correlated microscopic surface imaging were obtained from six swine hearts. A total of 61 volumes were obtained, where 34 volumes were from the right ventricle and

27 from the left ventricle. An additional two hearts were used for evaluation of catheter based imaging of epicardial substrates

119

The myocardium is covered by a thin layer, called the epicardium, which appears highly reflective within OCT images. The normal myocardium had a characteristic birefringence artifact, due to the highly organized structure of fibers in the myocardium

(Figure 4. 3. Epicardial fat has a heterogeneous appearance within OCT images (Figure

4. 4). Adipose tissue is covered by a layer of epicardial cells and connective tissue that appears as a bright layer within OCT images. Coronary vessels appear as signal poor regions, corresponding to the empty vessel lumens (Figure 4. 5) embedded in a layer superficial to the myocardium. The location of the vessel lumens correlate well to the location of the vessels apparent in the microscope images. The unique characteristics of three tissue types are shown in Figure 4. 6, where OCT volumes were recorded in heterogeneous regions with mixed tissue types. Figure 4. 6(a) represents an area of fat surrounding a coronary blood vessel and Figure 4. 6(b) represents an area with coronaries, fat and myocardium. The distinct features of the epicardium, myocardium, epicardial fat and coronaries are visible within slices parallel to the epicardial surface, and correlate to the microscope images of the surface. Within the B-scan images, a thick epicardial layer is observed which covers epicardial fat, which in turn surrounds the coronary vessel (Figure 4. 6 e, g). A second volume encompassed an area with epicardial fat, coronary, and untreated myocardium (Figure 4. 6-c). Within the volume, there is a transition of B-scan images encompassing untreated myocardium characterized by polarization artifact and epicardial layer and a coronary vessel (Figure 4. 6-f) to B-scans encompassing epicardial fat, which appears heterogeneous, epicardium and coronary

(Figure 4. 6 6-h).

120

121

122

123

With the application of RF energy and lesion formation, the contrast between the

epicardium and myocardium and the polarization dependent artifact is lost. This is

shown in Figure 4. 7 where a volumetric image is obtained of an area encompassing a

portion of treated tissue and untreated tissue.

124

125

A view of the epicardial surface was obtained with the forward scanning probe in air and in the presence of saline. As shown in Figure 4. 8, catheter based OCT images of epicardial structures show similar image features as those seen in OCT images obtained with the bench-top scanner, but with slightly poorer lateral resolution. The epicardial layer, and polarization artifacts due to the birefringent myocardium are visible within catheter-based images. Coronary vessels were also visible within catheter-based images. With a decrease in lateral resolution, there is an observed difference in the textural appearance of epicardial fat within catheter-based images. The texture is still heterogeneous, however, and readily differentiable from myocardium.

126

4.3.1. Discussion

Currently, there is no direct visualization of the epicardium available to electrophysiologists performing epicardial RFA procedures. Here, we demonstrated that

127

OCT has the potential for guidance of epicardial radiofrequency ablation therapy by identification of coronary vessels, epicardial fat, and normal myocardium.

4.3.1.1. Substrate Characterization

Previous studies have characterized the optical properties of myocardium[162,

194], blood[194], epicardial fat[195], vessel wall of different compositions[194], and myocardial radiofrequency ablation lesions[162] and myocardial laser ablation lesions[195]. It was shown that epicardial fat has significantly lower absorption than myocardium[195], blood has high absorption[194], and myocardial RFA lesions have an increased scattering coefficient[162, 195]. Histological analysis of myocardium exposed to thermal damage showed decreased in birefringence within ablation lesions[196].

These varied optical properties are recapitulated at 1310 nm wavelength and provide intrinsic contrast within OCT images of epicardial substrates. In addition to bulk optical properties, each tissue type has characteristic microstructure that provides contrast within

OCT images.

The normal myocardium was characterized by a strong birefringence artifact and bright epicardial layer. These result are consistent with our previous results, imaging lesions created on the endocardial surface[186]. The loss of contrast between the epicardium and myocardium within an ablation lesion may be due to the increase in scattering coefficient of the necrotic myocardium, reducing the scattering mismatch between the epicardium and myocardium. Although the polarization-dependent artifact is strong, future work will explore the use of polarization-sensitive OCT for quantitative assessment of tissue birefringence.

128

Our experimental samples were freshly excised swine ventricular wedges, without coronary blood flow. We were able to identify vessels as signal poor regions, correlating to the lumen of the vessel. OCT has been demonstrated to visualize blood vessels in vivo in ophthalmology with the use of the Doppler effect and observing shadowing created by light absorption by blood[188, 197]. In vivo, coronary vessels will obviously be filled with blood, so the lumens will not appear as voids. However, shadows will be apparent due to light absorption and scattering by the blood. In addition, the use of Doppler

OCT[197] and/or speckle modulation[198] can be used to identify blood vessels using blood flow as the contrast mechanism. Therefore, we believe OCT will be able to identify coronary vessels of different sizes, by identifying tissue regions with a shadow and exhibiting flow. Epicardial fat has a clear, characteristic heterogeneous appearance within

OCT images. These results are consistent with pervious application of low coherence interferometery of adipose tissue from the breast, showing a heterogeneous axial scan profile[189].

4.3.1.2. Real time Guidance

Because of the severe adverse complications that can arise due to epicardial ablation, preliminary studies have been conducted to evaluate the use of a flexible, steerable endoscope, 5.3mm in diameter to guide epicardial procedures by identifying coronaries[23]. Air was used to expand the pericardial space and provided visualization of critical structures, RF catheter and ablation lesions. However, image quality was diminished in the presence of saline flushing. As compared to the use of an endoscope for guidance, OCT provides subsurface imaging and can be implemented in small diameter catheter probes. As demonstrated here, even with 30µm lateral resolution,

129

catheter-based OCT can be used to identify critical structures. Imaging with a catheter probe, the same heterogeneous pattern was observed with images of epicardial fat, identification of vessel lumens, epicardial layer and polarization artifact.

Due to the fact that there is no convective cooling of the tissue surface during epicardial ablation, as there is during endocardial ablation due to blood flow, the use of saline irrigated RF catheters are becoming increasing common. Saline irrigated catheters keep the surface of the tissue cool, allowing increased power delivery, resulting in deeper lesions. Real time viewing with the prototype flexible forward imaging OCT catheter provided a visualization of the epicardial surface. Maintenance of adequate tissue contact is one of the main parameters ensuring efficient energy delivery. Using the forward imaging probe, tissue contact can be assessed in the presence or absence of saline flushing. OCT has been demonstrated to image in the presence of saline flushes for intravascular imaging[86] and within a saline bath with super-perfusion flow. Although the field of view of OCT is limited, real-time imaging with OCT, integrated with RF ablation can potentially inform the operator when the ablation catheter is in the vicinity of epicardial fat or a coronary vessel.

4.3.1.3. Future Work

All of the current work was conducted using freshly excised swine hearts. We demonstrated that with OCT, we can observe similar image features in OCT images of epicardial fat, normal myocardium, and coronary vessels using bench-top and catheter- based imaging. Future work is needed to evaluate the ability for OCT to observe these features in large animals in vivo. These experimental conditions will introduce motion of a beating heart and blood flow within coronary vessels. Other epicardial substrates of

130

interest are myocardial infarction[199] for the ablation of ventricular tachycardia and the pulmonary veins [200], during epicardial ablation of atrial fibrillation.

4.3.2. Conclusion

This study demonstrates that optical coherence tomography has the potential for guidance of epicardial radiofrequency ablation procedures. Our experimental results show that critical epicardial structures, coronaries, epicardial fat, and myocardium, have distinct appearances within OCT images. A direct image by OCT can aid in guiding the application of energy, by identifying coronary vessels to determine if energy delivery can be carried out safely. In addition, confirmation of epicardial fat will reduce ambiguity in interpreting electrograms and will allow appropriate titration of RF power dosage or increased energy delivery duration to increase lesion efficacy. Together, guidance by

OCT to provide real-time cross sectional imaging for substrate identification may decrease complications and increase success rates.

131

Chapter 5: Real-time Monitoring of Cardiac Radiofrequency Ablation Lesion Formation using an Optical Coherence Tomography Forward Imaging Catheter

5.1. Introduction

Current techniques to guide ablation therapy utilize low resolution two dimensional fluoroscopic images or static images from computed tomography merged onto fluoroscopy. The monitoring of successful formation of an ablation lesion is by indirect means, such as on line assessment of tissue temperature, power delivery, and impedance at the tip of the catheter. This limited, indirect method of monitoring during ablation procedures may often result in delivering more ablation lesions than necessary to achieve a therapeutic effect, prolonging procedure times, thereby increasing the risk of these procedures. The current implementation of RFA in clinical practice is associated with significant unmet technological needs, since there are no direct measures of the successful delivery of ablation lesions. Imaging has important roles in improving the

132

safety and decreasing the time of RFA procedures by providing an anatomical map of the heart and pulmonary vessels with computed tomography (CT) or magnetic resonance imaging (MRI) [201], by evaluating postprocedural effects of ablation lesions using

MRI[17], and by evaluating tissue-electrode contact within the pulmonary veins using echocardiography[20]. All of the above imaging techniques require offline assessment before and or after RFA. Additional technology for directly monitoring ablation lesion formation during procedures in real-time may further decrease procedural time and improve patient and operator safety.

OCT has demonstrated the ability distinguish thermal ablation lesions created with radiofrequency energy from untreated tissue[186](Chapter 3). Viable tissue has a banding appearance within OCT images due to the birefringence property of the highly organized myocardial tissue. Ablated tissue loses visible fiber structure within OCT images[186] and its birefringence properties[57]. OCT has been demonstrated to image purkinje network[87], imaging the fast and slow pathways in the atrial-ventricular (AV) node[31, 88], and abnormal fiber organization[31, 89], targets for RF ablation. These targets for ablation can be readily distinguished within in vitro animal models.

The clinical utility of OCT monitoring of cardiac RFA therapy includes providing a direct visualization of the catheter-heart interface, direct visualization of normal areas of the heart to avoid during ablation, and direct visualization of lesions that are made.

Together this may reduce procedural time, with guidance for better placement of critical lesions for ablation, and improve patient and operator safety with reduction in radiation exposure. We hypothesize that OCT can provide real-time, high resolution, subsurface imaging to monitor formation of ablation lesions in real time.

133

5.2. Methods

5.2.1. Feasibility Testing

To test the feasibility of using OCT to track dynamics due to RF energy delivery and visualize lesion formation, ablation lesions were created on bovine ventricles. An

EPT-100XP generator (Boston Scientific) was used to generate lesions on the endocardial surface of the right ventricle under a constant power protocol. The ablation catheter was positioned such that the catheter was visible in the margins of the OCT image.

Figure 5. 1. Experimental setup for imaging real time dynamics of ablation lesion formation using a bench top OCT scanner.

OCT images were acquired at eight frames per second. This frame rate was sufficient to test feasibility as clinical lesions are generated over the course of 60 to 120 seconds. Representative images from these experiments of visualizing dynamic changes due to RF energy delivery are shown in Figure 5. 2. After approximately two seconds, a

134

lesion begins to develop at the margins of the ablation catheter. As shown in Error!

Reference source not found.C, image intensity increases as a function of time.

Within a second demonstration of imaging real time dynamics, over the course of the energy delivery, an increase in image intensity was observed by the margins of the catheter. Thereafter, debris formation was observed, which we believe to be coagulum at the margins of the RF catheter. After 4 seconds of energy delivery and a high impedance of 312Ω, a steam pop occurs. The catheter was removed and the sample was shifted by 3 mm after showing visible craters and cardiac perforation at the center of the lesion. The coagulum observed in 3.5 seconds after the start of energy delivery (Figure 5. 3C) may be a precursor to the development of significant craters, and feedback to stop energy delivery or titrate dosage.

135

Within the seven lesions created under this experimental protocol, all resulted in steam pops within the first 10 seconds of energy delivery. When imaging real time dynamics of RF energy delivery with the long working distance OCT scanner, the ablation catheter was not submerged in a bath of PBS with superfusion flow. Imaging with a sample submerged in a bath with two cm of PBS resulted in decreased image quality and intensity due to water absorption. Therefore, to adequately demonstrate imaging of dynamics due to RF energy delivery, a catheter with a short working distance that can image in the presence of solution flow is needed.

OCT catheter probes development has been used extensively for gastroentestinal endoscopy and intravascular imaging[75, 202]. A variety of designs have been employed including forward imaging[63-66], sideviewing probes[67], and both forward and sideviewing probes[68]. Designing a forward imaging probe that is both small in diameter and flexible presents many challenges. Many have designed probes with proximal actuation to provide a wide range of scanning patterns of the beam at the tip of the catheter. However, these probes are not easily miniaturized, as the scanning

136

mechanism limits the outer diameter of the probe tip. Distal actuation has been shown in the paired angle probe[64], but requires a ridge probe.

5.2.2. Catheter Designs

Cardiac RFA lesions are formed most effectively if the catheter is perpendicular to the tissue surface and has adequate contact. Therefore, a flexible forward scanning

OCT catheter was designed and prototyped that allows for contact imaging. Forward imaging probes were designed with increasing complexity and capabilities, using the optical design software package ZEMAX, axial image only, circular scanning, and arbitrary scanning pattern (including linear, circular, and volumetric). An axial line can be achieved using a grin lens. To obtain circular scanning, a grin lens glued to a wedge prism. Varying the angle of the wedge prism, an increase in lateral scanning range is possible. To obtain arbitrary scanning, a gradient index lens and two angled BK7 glasses are used. When the grin-angled glass pair is rotated and the second angled glass pair is stationary, we can obtain circular scanning. If the second angled glass pair is rotated in an equal but opposite angle, linear scanning is possible. Figure 5. 4 summarizes the

Zemax simulations for the axial and circular scanning designs. By observing the point spread function of the two optical systems, we can see that both maintain a 30µm FWHM spot size, however the circular scanning probe has an increase in aberrations.

137

The final design for the forward imaging probe provided circular scanning, where the prism was rotated to reduce aberrations. The design specifications of the catheter are summarized in Table 5. 1. Parameters are related to optical, mechanical and

138

environmental performance was developed to ensure that the prototype catheter can be used for monitoring in real-time dynamics due to RF energy delivery. Briefly, a short working distance catheter that can provide contact imaging in the presence of physiological solution flow (blood or PBS) was necessary. For the initial prototype, the catheters had a desired 2mm scan diameter, and potentially accommodate a broadband

1310nm ∆λ140nm light source.

Table 5. 1. Forward imaging OCT catheter design specification for optical characteristics, mechanical characteristics, and environmental conditions

139

The basic principles of the probe and how to interpret the images obtained from circular scanning are shown in Figure 5. 5-a. The catheter uses a GRIN lens to focus the light and a Risley prism to deflect the beam off axis. Risley prisms are used commonly for beam steering and as a mechanism to increase the field of view of standard endoscopes. By rotating the entire optical assembly, circular scanning is obtained. This results in an image whose length at the top is the circumference of the scanning profile.

The prototype catheter design in shown in Figure 5. 5-b. Light is delivered to the end of the catheter via a SMF28e optical fiber, where the beam is focused by a GRIN lens, deflected 1 mm off-axis by a Risley prism, and a fused-silica optical window isolates the probe interior from the tissue environment (Figure 5. 5A).

The mechanical support was designed entirely from glass, ceramic, and polymer materials (no metal) to avoid unwanted thermal effects and interference in close proximity to radiofrequency ablation catheters (Figure 5. 5b). The end cap was made out of polyetheretherketone (PEEK), a rigid plastic that can maintain its mechanical properties up to 250oC. Rotary motion imparted by torque applied to tight-buffered fiber at the fiber rotary joint allowed for circular scanning. The rigid portion of the catheter was 18mm in length and the outer diameter was 2.5mm. Polytetrafluorethylene (PTFE) moisture seal heat shrink tubing was placed on the probe end cap, increasing the outer diameter at the probe tip to 3.2mm. The probes, which are 2 meters long, are axially flexible for manipulation within the heart, but torsionally rigid to provide circular scanning and the protective outer sheath is biocompatible. Eight catheters were prototyped with this design. The mean and standard deviation of critical parameters of the

140

prototype probes was as follows, scan diameter 2.05 (0.1) mm, rigid end length 20.8 (1.1) mm, insertion loss 0.16 (0.1) dB, and 33.6 (2.8) dB return loss.

The spot diameter of the design is 28 µm full width at half-maximum (FWHM), with minimal aberrations over more than 1 mm of axial scan range from the optical glass, simulated with a broadband light source centered at 1310nm with a 140nm bandwdith

(Figure 1c). To ensure that the probe maintains a 30µm spot size with minimal aberrations within a temperature range experienced in vivo and in close proximity to ablation catheters, simulations using the optical design software package ASAP were conducted over a range of temperatures. The grin lens is described by the parameters no and A1/2, and diameter. The index of refraction within the GRIN lens varies in the radial

1/2 direction, nr, as shown in Equation 11. The A parameter was varied based on manufacturers specifications to compute a new index profile as a function of temperature.

The spot profile was simulated at 20oC, 70oC and 95oC.

⎛⎞A 2 Equation 11 nnro=−⎜⎟1 r ⎝⎠2

Were no is the index of refraction along the optical axis and r is the radial distance.

141

Figure 5. 5. OCT Forward Imaging Probe. a) Schematic showing basic principles of optical design of probe for circular scanning, grin lens used for focusing, risley prism to deflect beam off axis, and optical glass to isolate probe from environment. Rotation of entire assembly produces circular scanning. b) Mechanical design of distal end of probe. c) Spot profiles at varying axial depths from ASAP simulation. d) Spot profile from prototype catheter. e) Close up view of proximal end of forward imaging probe. RJ – ring jeweled bearings, GR – Grin lens, RP – risley prism, OG – optical glass, FB – fiber, FE – ferrule, EC – end cap, S – sheath The forward imaging catheters were integrated into a Fourier domain optical coherence tomography (FDOCT) system, Error! Reference source not found.. A superlumincent diode centered at 1310 nm with a 73 nm (FWHM) bandwidth was used for the light source (Inphenix). A linear in wavenumber (k=2π/λ) spectrometer [50] was used to project spectral interference fringes onto a 1024-pixel InGaAs line-scan camera

(Goodrich) capable of more than 40,000 A-scans per second. The spectrometer has a 4.3 mm imaging range, 2mm -6dB fall off, and 115dB signal to noise ratio. The axial resolution was measured by obtaining the intensity profile of the optical glass within the probe sheath. The measured axial resolution was 15µm.

142

Representative images from the forward imaging probes are shown in Figure 5. 7.

The image quality varied by probe. Non-uniform rotational rates due to high friction resulted in streaked appearances within images from a subset of probes. This is a common problem observed in intravascular imaging catheters, called non-uniform rotational distortion (NURD). Probes CW06 and CW07 produced the best quality images and were used for subsequent experiments.

143

Feasibility of visualizing real-time ablative lesion formation using the forward imaging catheter was demonstrated ex vivo, using ventricular wedges from a freshly excised swine heart. Following onset of general anesthesia a lateral thoracotomy was performed and the heart was rapidly excised and placed in ice cold phosphate buffered saline (PBS). Individual sections of ventricular muscle were placed in a custom chamber with PBS maintained at 37oC. Ablation lesions were created with a temperature controlled (80oC) protocol with a maximum delivered power of 50W using the Maestro

3000 generator (Boston Scientific). Endocardial lesions on the left ventricle were created

144

using an 8Fr, 5mm tip catheter Blazer II (Boston Scientific). RFA energy delivery was delivered for 60 seconds.

5.2.5. Imaging Protocol

The OCT forward imaging probe was bound side-by-side to the RFA catheter

(Figure 5. 8a). Torque was applied to the catheter by a rotary joint. An encoder placed on the rotary joint provided a frame sync signal for every revolution of the rotary joint.

During the experimental protocol, imaging was conducted at 20 frames per second with

2000 A-lines per frame. Baseline imaging was conducted for 15 seconds prior to the start of RF energy delivery and 15 seconds after the conclusion of RF energy delivery. The effect of the presence of blood was evaluated by imaging control sites after being submerged in a bath with heparinized swine blood.

5.2.6. Validation Staining with triphenyltetrazolium chloride (TTC) was used to validate lesion formation [172]. The tissue was sliced through the center of the lesion and incubated in

1.0% TTC in PBS for 30 minutes at 37oC.

145

5.2.7. Image Analysis

To correct streaking artifacts due to non-uniform rotational rates, a correlation based method previously described was used[193]. Adjacent A-scans that are highly correlated are assumed to result from regions in which the catheter was rotating too slowly. Therefore, a new image was created by removing highly correlated A-scans.

The first A-scan of the new image is the first A-scan of the original image. The correlation coefficient is computed between the first A-scan and the second A-scan. If the correlation coefficient falls below a threshold, the A-scan is added to the new image.

Otherwise, the third A-scan in the original image is compared to the first A-scan. Once an A-scan falls below a threshold, it is added to the new image and because the new A- scan to compare subsequent A-scans to. Figure 5. 9 shows an example of the correlation based method to reduce streaking artifacts within images.

146

Figure 5. 9. Image visualization of catheter images. a) Forward imaging probe uses GRIN lens for focusing and Risley prism for beam deflection. b) OCT image obtained by forward imaging probe. c) Zoomed in region from OCT image in (b) showing areas with streaking. d) OCT image after correlation based algorithm for correcting non-uniform scanning. e) Zoomed in region from OCT image in (d) showing removal of streaking artifact. f) Decimated version of corrected OCT image from (d) to be used for image analysis. Example averaged axial scan is plotted in red. g) Visualization of OCT image from (d) displayed with correct aspect ratio. Image acquired with CW07 and FDOCT system. After image correction to remove highly correlated A-lines, image analysis parameters were extracted to track dynamic changes in optical properties due to RF energy delivery. Preprocessing was conducted to normalize the intensity of each image

147

and remove the noise floor. Each image was decimated to 512 x 20pixels. The region of interest (ROI) analyzed was 370 microns in length in the axial dimension, starting 100

µm below the sample surface to avoid the endocardial layer. A constant index of refraction of ventricular muscle, 1.382[170], was assumed for all tissue samples. Mean intensity within the ROI, and convolution with a 60 x 1 Laplacian of Gaussian kernel to compute the gradient associated with birefringence banding. The software package

MATLAB was used to implement the image analysis algorithms.

5.2.8. Statistical Analysis Analysis of variance (ANOVA) was conducted to test when image analysis parameters become statistically different from baseline values. Parameters were divided into six groups, increments of 15 seconds during the experimental procedure, as shown in

Figure 2-b. The software package Origin was used for statistical analysis. A p-value less than 0.05 were considered statistically significant. All results are reported as mean (1 standard deviation).

5.3. Results

5.3.1. Forward Imaging Probe A forward imaging probe was prototyped with no metal that provides contact, circular imaging with 30µm lateral resolution and a 2mm scan diameter. Spot profiles from a representative probe using a SLED centered at 1310nm and bandwidth of 75nm

(Inphenix) (Figure 5. 5-d) demonstrates that the probe produces a circular spot profile under 30µm. Minimal aberrations were observed which compared favorability to simulations using the optical design software ASAP. Simulations of the spot size for varying temperature showed that the FWHM was less than 30µm for all cases expect at

148

95oC, 1mm from the optical glass. At 95oC, the FWHM increased slightly to 30 µm and

31 µm for the x and y directions respectively 1mm from the optical glass.

FWHM(µm) at distance from optical window

GRIN Lens Temperature Axis 0mm 0.5mm 1mm

37oC x 29 28 29

y 28 28 30

75oC x 28 28 29

y 27 28 30

95oC x 28 28 30

y 27 28 31

Table 5. 2. Spot size as a function of GRIN lens temperature

Representative images taken with the forward imaging probe at 10 frames per second with 4000 A-lines per frame are shown in Figure 5. 10. Figure 5. 10-a is an image of in vivo human thick skin in which the layer structure and sweat ducts are visible. Images were also acquired of critical epicardial substrates, coronary vessels that have to be avoided during epicardial ablation.

149

5.3.2. Real time visualization of lesion formation

Baseline images, from the first 15 seconds of the experimental protocol, was characterized by a strong birefringence band and relatively low signal intensity within untreated tissue. Convolution with a Laplacian of Gaussian was used to highlight the intensity change due to the birefringence band. The variability of the gradient strength and mean intensity was small during the baseline period, 0.039(0.002) and 0.45(0.006) respectively. With the application of RF energy, there was a significant decrease in gradient strength within the first 3 seconds. A significant increase in image intensity was observed after 15 seconds. Mean intensity and gradient strength showed a nearly linear change with RF energy delivery (Figure 5. 11-g,h,i).

150

151

152

A second demonstration of imaging real time dynamics is shown in Figure 5. 12.

Baseline images show two birefringence bands. With the application of RF energy, we can visualize the bands broadening and appearing deeper into the tissue. A significant increase in signal intensity was observed within the first 5 seconds of RF energy delivery, and thereafter increased nearly linearly with RF energy duration. There was also a significant decrease in gradient strength observed within 4 seconds after the start of RF energy delivery.

In both demonstrations, mean intensity provided a nearly linear change with RF energy delivery. Mean intensity and gradient strength were highly correlated, with a 0.89 and 0.60 correlation coefficient for examples one and two respectively. However, gradient strength had a -73% and -167% change between baseline (0-15 seconds) and post-ablation (75-90 seconds) in the two example lesions. In contrast, mean intensity showed a 22% and 13% change.

5.3.3. Imaging in the presence of blood

During RF energy delivery, adequate catheter – tissue contact is necessary for energy delivery. While imaging in the presence of blood, an image of the myocardium was obtained when the catheter was in direct contact with the tissue, displacing the blood

(Figure 5. 13). Blood strongly attenuates light at 1310nm. When the catheter probe was not in direct contact with the tissue, the mean imaging depth was significantly reduced, p<0.001, Figure 5. 13-b.

153

5.4. In vivo imaging using the forward imaging probe In the previous section, we presented the first demonstrations of real time visualization dynamics due to RF energy delivery in cardiac tissue. The results from these experiments shows the promise that OCT technology can be used for monitoring intra cardiac ablation procedures and encourages the next step of feasibility testing, in vivo imaging. It is our objective to demonstrate introduction of the catheters into the heart and investigate and develop techniques to solve two main technical problems associated with imaging the endocardium. First, blood strongly attenuates 1310nm OCT probe light, so a significant quantity of blood should not be in the path of OCT imaging.

Second, as the heart wall is not stationary, high-resolution OCT imaging is sensitive to even small motions. However, since the catheter is designed to be in contact with the myocardium during the ablation procedure, we anticipate that motion and imaging through blood will not be problematic. Blood will almost completely be displaced by contact of the probe with the tissue. With video rate imaging, the manual stabilization of the RFA catheter by the operator should be sufficient to mitigate significant motion artifact in the OCT imaging. Presented below, to our knowledge, is the first

154

demonstration of in vivo intra-cardiac imaging using OCT. We have demonstrated that real time imaging with a forward imaging catheter can assess tissue-catheter contact, visualize critical structures, and visualize dynamics due to RF energy delivery.

Both epicardial and endocardial imaging was conducted with the forward imaging catheter in an open chest procedure in a female swine. The OCT forward imaging catheter was strapped side by side to the RFA catheter as shown in Figure 5. 14-A. The catheters were inserted through purses strings created in the right atria (Figure 5. 14-B).

After endocardial imaging, imaging was conducted on the epicardium (Figure 5. 14-C).

All endocardial imaging and ablation was conducted on the right side of the heart. All animal studies was conducted according to protocols approved by the Institutional

Animal Care and Use Committee of CWRU.

Figure 5. 14. In vivo cardiac imaging in a swine. a) OCT forward imaging probe (green arrow) and RFA catheter (yellow arrow) were strapped side by side for preliminary evaluation of imaging dynamics due to RF energy delivery. b) The in vivo experiment was conducted in an open chest procedure where the combined OCT and RF catheters were inserted into the right atria. c) Epicardial imaging was also conducted with the combined catheters.

The OCT probe was integrated into a FDOCT system with linear-in-k spectrometer. All imaging was conducted with OCT probe CW06 at 20 frames per second, with 2000 lines per frame. The RF/OCT catheter was advanced and navigated

155

within the heart under fluoroscopic guidance. Hemodynamic parameters remained relatively constant throughout the experimental procedure (4 hours), demonstrating initial safety of endocardial OCT imaging with the forward imaging probe. Images of the endocardial surface was obtained when the catheter was in direct contact with the surface wall. Figure 5. 15 are example images in the right atria, when the catheter was not in contact (Figure 5. 15-a) and in contact (Figure 5. 15-b) with the endocardial surface. In vivo images recapitulate results observed within bench-top imaging, ex vivo experiments.

A layered appearance and textures observed within some areas of the atria were similar to ex vivo experiments (Figure 3. 18).

156

Birefringence dependent bands were observed within a subset of images of normal tissue. two examples of the right ventricule epicardium nad right atria Endocardium are shown in Figure 5. 16.

Figure 5. 16. Observance of birefringence dependent bands in vivo in the A) right ventricle epicardium and the B) right atria endocardium

To evaluate if OCT can observed dynamics due to RF energy delivery in vivo, RF energy was delivered for 60 seconds with a temperature controlled and power controlled protocols. RF energy was delivered with a Maestro 3000 generator (Boston Scientific) and lesions were created using an 8Fr, 5mm tip catheter Blazer II (Boston Scientific). The imaging protocol followed that of the ex vivo real time imaging experiments. A baseline

15 seconds of imaging was recorded once stable contact with the endocardial surface was made. A steam pop was occurred during a temperature controlled protocol, where the target temperature was 85oC. Figure 5. 17 shows images from the time series, where we believe coagulum is being developed, and then the progressive increase in size of voids within the myocardium.

157

5.5. Discussion

We present a flexible forward imaging OCT catheter made without metal that provides circular scanning by distal actuation of the fiber. Using a forward imaging catheter, OCT can provide real-time direct visualization of RFA therapy, assessing tissue

– electrode contact, confirming energy delivery, visualizing lesion formation, identifying critical epicardial structures and imaging in the presence of blood.

5.5.1. Forward Imaging Probe

The prototype forward imaging catheter probe demonstrated imaging at 20 frames per second in a variety of environments. The probe maintains a small circular spot size over a large range of temperatures that will be experienced in vivo and during monitoring

158

of ablation therapy. Real time imaging was conducted with the probe submerged in a bath with phosphate buffered saline maintained at 37oC with super perfusion flow and placed in contact with freshly excised ventricular wedges from swine hearts, demonstrating that imaging can be conducted within physiological solutions with flow. Although simulations showed a small increase in spot size at 95oC, this is the temperature in which ablation generators stop energy delivery to prevent coagulum formation and represents an extreme case.

For a forward looking catheter, circular scanning is a not a standard scanning pattern used to generate OCT images and may take a while to get use to. However, circular scanning has many positive attributes. With contact imaging, the field of view is limited by the clear aperture of the optical glass. Therefore, with traditional linear scanning, the lateral scanning range is limited to the diameter of the clear aperture of the optical glass. Circular scanning allows for a large lateral imaging range, where the limit has increased to the circumference of the clear aperture of the optical glass. With our designed 2mm scan diameter, our lateral imaging range is 6.28mm. In addition, when imaging a sample within an optically clear fluid or air, contact angle can be assessed easily with circular scanning.

The design of this catheter was conducted for intra cardiac electrophysiology, where probes are disposable and procedures last 2-8 hours. Future designs can produce probes that are durable over a longer range of time. Our current prototype probe is longer and more flexible than standard RFA catheters. Therefore, increasing the stiffness of the fiber buffer and decreasing the fiber length will increase the torsional rigidity of the probe and increase its lifespan.

159

This design can be adapted to many applications. The distance between the optical glass and the Risley prism can be changed for a forward imaging long working distance probe. Since the rotary mechanism is imparted by torque on the fiber, this probe can be easily miniaturized with the use of custom designed parts. Lastly, the probe was designed entirely without metal to prevent unwanted interference with the RF electrode.

This has other clinical potential including being a MR compatible OCT catheter.

5.5.2. Real Time Visualization of Lesion Formation We demonstrated that OCT has the potential to monitor in real time the formation of cardiac ablation lesions using a flexible forward imaging catheter. Real time visualization of RFA therapy with OCT showed changes in tissue birefringence within the first 3 seconds of RF energy delivery. A decrease in gradient strength and increase in intensity were the two parameters that showed a linear change with RF energy delivery within all examples. These results are consistent with previous work showing an increase in intensity and decrease in tissue birefringence due to thermal ablation. Imaging with polarization sensitive OCT (PSOCT) would allow for quantitative estimation of tissue birefringence. PSOCT would remove polarization dependent artifacts within images of untreated samples, which would allow for quantitative assessment of attenuation coefficient from OCT images, which may provide another analysis parameter that tracks with RF energy delivery time.

5.5.3. Assessment of Tissue Contact in the Presence of Blood

In addition to visualizing lesion formation in real time, OCT has clinical utility for assessing tissue – RFA electrode contact. The optical glass within the probe provides a strong, fixed signal within images and can be used as a reference for assessing tissue

160

contact. Assessment of tissue contact is further enhanced in the presence of blood.

1310nm light is strongly absorbed and scattered by blood. When the catheter is not in direct contact with the tissue surface, the imaging depth is significantly reduced. This significant decrease in imaging depth is a potential feedback mechanism to ensure and maintain contact during RFA procedures. One of the most important factors that affect lesion size is the maintenance of tissue – RFA electrode contact. If the catheter is not held with sufficient pressure to the endocardial surface to maintain a stable contact during the delivery of RF energy, motion and the inclusion of blood in the OCT field of view, will be directly realized by a significant decrease in imaging depth as shown in Figure 6.

During intravascular imaging using OCT, saline purges or balloon occlusion is used to eliminate blood from the field of view during OCT imaging. RFA saline irrigated catheters are often standard tools during complex RFA procedures. These catheters continuously flush small amounts of saline at the tip of the RFA catheter to cool down the endocardial surface during RF energy delivery. Making contact with an OCT catheter during RFA lesion formation in combination with saline irrigated catheters will further reduce blood present in the OCT field of view.

5.5.4. In vivo imaging We demonstrated the first in vivo OCT intra-cardiac imaging. This was enabled by using a forward imaging probe with a short working distance. A view of the endocardial surface was only obtained when the catheter was in direct contact. Real time dynamics due to RF energy delivery were also observed. However, the birefringence dependent band observed within ex vivo experiments were not consistently observed within in vivo experiments. They were only observed within a few images imaging the right atria from the endocardial surface and the right ventricle, imaging from the

161

epicardial surface. We speculate that motion during the cardiac cycle has a significant effect, and optical properties are changing during the cardiac cycle. Therefore, future implementations to observe the birefringence band or to quantify birefringence using

PSOCT will require imaging speeds greater than 20 frames per second. Within our current experiments, our image was significantly oversampled. With 2000 lines per frame, 6.3mm lateral field of view, each pixel was 3.15µm. The spot size of the forward imaging probe is 30µm, and we are currently oversampling the spot by almost a factor of

10. 700 lines per frame will result in a 9µm spot size, and an imaging speed of 57 frames per second. Future mechanical analysis is needed to identify a fiber buffer that can withstand torque generated by rotating the catheter at speeds between 50 – 100 frames per second over 12 hours.

These results show that imaging with a forward scanning OCT catheter can assess tissue-catheter contact and visualize dynamics due to RF energy delivery, and visualize textures within critical structures. Future studies can be conducted to address questions such as predicting over-treatment, providing feedback to titrate RF energy delivery, and guiding RFA procedures by identifying anatomical targets for ablation and normal structures to avoid. Real-time monitoring and guidance may help to improve the efficacy and reduce procedure time within complex ablation procedures such as isolation of the pulmonary veins for the treatment of atrial fibrillation or ablation of surviving viable tissue within an infarct border zone for the treatment of ventricular tachycardia. The time course of steam pop development was fast. Therefore, it will be necessary to develop and automate an algorithm for detecting precursors to steam pops. This will

162

require further in vivo experiments with correlated functional measurements to determine if OCT can determine precursors to steam pops in enough time before craters are formed.

5.6. Conclusion

In summary, catheter based optical coherence tomography can be used for real time monitoring of ablation lesion formation and imaging in the presence of blood using a forward imaging probe. A direct image by OCT has the potential to guide the precise application of energy, avoid normal cardiac structures where ablation could be harmful, provide real-time formation of successful lesions and assess contact. Importantly, this may decrease the procedure time and radiation exposure to the patient and physician.

163

Chapter 6: Summary and Future Work

6.1. Summary

Optical coherence tomography (OCT) is an emerging imaging modality that provides depth resolved, high resolution, images with high acquisition speeds. Within this dissertation, we developed OCT technology for applications in cardiac electrophysiology. There is currently limited real time imaging tools that provide high resolution (~10µm) images, which is needed to observe early stage structural changes in the myocardium due to disease or therapy.

Within Chapter 2, we explored the use of OCT to visualize and characterize arrhythmogenic substrates. Abnormal changes in fiber orientation or fiber disarray increase the likelihood of arrhythmia. We presented optical coherence tomography

(OCT) as a method to image myofibers in excised intact heart preparations. An automated algorithm for fiber orientation quantification in the plane parallel to the wall surface was developed. Quantifying fiber orientation in the plane parallel to the wall surface from OCT images can be used to help understand the conduction system of the specific sample being imaged. OCT has the potential for identifying substrates for

164

ablation. Abnormal fiber orientation caused by a disease such as infarction or hypertrophic cardiomyopathy is a substrate for arrhythmia that may result in sudden cardiac death. Volumetric images were acquired of healed myocardial infarctions in small animal models, rabbits and mice. Ventricular wall thinning and increased intensity was observed within the mouse models.

Radiofrequency ablation (RFA) therapy is now the standard of care for the treatment of cardiac arrhythmias. Current techniques to guide ablation therapy utilize low resolution two dimensional fluoroscopic images or static images from computed tomography merged onto fluoroscopy. The clinical utility of optical coherence tomography (OCT) monitoring of cardiac RFA therapy includes providing a direct visualization of the catheter-heart interface, direct visualization of normal areas of the heart to avoid during ablation, visualization of areas to target for ablation and direct visualization of lesions that are made. Within Chapter 3, ex vivo experiments using a wedge swine model were conducted to test the hypothesis that OCT can distinguish necrotic from viable tissue, identify signs of overtreatment (steam pops and craters), and evaluate the dimensions of the lesion. Image analysis tools were developed that distinguished ablation lesions from untreated tissue within ventricular wedges with high sensitivity and specificity. Gaps of viable myocardium had a characteristic birefringence band between two ablation lesions and correlated with triphenyltetrazolium chloride staining to identify viability and necrosis. Over treated lesions were observed with characteristic disruptions within the myocardium and increased tissue heterogeneity.

Changes in optical properties and loss of birefringence provide intrinsic contrast between untreated sites and ablation lesions.

165

Within Chapter 4, we demonstrated the feasibility of OCT guidance for epicardial ablation procedures. OCT image features were observed that clearly distinguish untreated myocardium, ablation lesions, epicardial fat, and coronary vessels and assess tissue contact with catheter based imaging. The normal myocardium had a characteristic polarization artifact band, due to the highly organized structure of fibers in the myocardium. Epicardial fat had a heterogeneous appearance, and the adipose tissue was covered by a layer of epicardial cells and connective tissue that appears as a bright layer.

Coronary vessels appeared as signal poor regions, corresponding to empty vessel lumens.

These features were also observed within catheter based imaging. These results support the potential for real-time guidance of epicardial RFA therapy using OCT imaging.

Within Chapter 5, we demonstrated that OCT can visualize dynamics due to RF energy delivery and assess tissue contact in the presence of blood. A forward scanning

OCT catheter was prototyped that provides contact, cone scanning with no metal, 30µm lateral resolution and 2mm scan diameter. During the application of RF energy, the mean intensity increased nearly linearly with time and gradient strength had a significant decrease at the onset of RF energy delivery. Samples were thereafter submerged in heparanized blood, where an image of the myocardium was obtained when the catheter was in direct contact with the tissue, displacing the blood. A direct image by OCT has the potential to assess contact in the presence of blood and monitor in real time ablation lesion formation. Imaging of dynamics due to RF energy was demonstrated with the prototype catheter in vivo in a swine.

6.2. Clinical Significance

166

The market for therapeutic cardiac technologies is large and growing rapidly[203]. The Frost and Sullivan’s U.S. Cardiac Rhythm Management Markets 2006 report listed imaging techniques to improve radiofrequency ablation procedures as the second driver for the cardiac ablation market, 2006-2012 and refinement of ablation techniques to treat atrial fibrillation which will expand the number of ablation procedures as the number one market driver[203]. Real time monitoring and guidance has the potential to predict events of over-treatment, providing surrogate signals related to lesion size, providing feedback to titrate RF energy delivery, and guiding RFA procedures by identifying anatomical targets for ablation and normal structures to avoid.

OCT has the potential to be used in a wide range of ablation procedures for real time monitoring and guidance, as the catheter can be easily minturized and potentially integrated into the RF catheter. This integrated catheter will be able to answer critical questions that will provide integral information during the procedure for decision making.

1. Assessment of contact and contact angle

2. Confirm lesion formation

3. Detect early damage

4. Identify critical structures

Maintaining adequate contact with the surface is essential for efficient RF energy delivery. Heat produce is proportional to the inverse square of catheter-tissue distance.

Fluoroscopy and impedance are standard methods for assessment of contact and echocardiography in complex procedures. An integrated RF/OCT catheter will easily assess catheter contact by visualizing changes in imaging depth. With constant feedback on electrode-tissue contact, there will be increased efficiency in RF energy delivery.

167

Maintenance of catheter contact will also increase lesion depth and increase the number of lesions that are formed. This may have a direct impact on success rates, as it has been shown that larger lesions prevent arrhythmia recurrence. Once stable contact is maintained, OCT imaging will be able to visualize dynamics of energy delivery and validate that a lesion has been formed by observing changes in image intensity and tissue birefringence. Being able to validate lesion formation will reduce ambiguity before electrophysiology testing to confirm efficacy of lesion formation.

Cardiac perforation is one of the most serious complications related to RFA therapy. It is current practice to use the increase of tissue-electrode impedance as a signal to detect overtreatment of the tissue and coagulum development[6]. However, previous studies have shown that adverse effects such as steam pops are not always associated with a large change in impedance[175]. More than ever, this is important for saline irrigated RFA catheters. Real time monitoring using OCT imaging can identify early signs of complications, such as voids within the myocardium, precursors to cardiac perforations. Lastly, OCT can potentially be used for procedural guidance. Guidance targeting arrhythmia substrates, such as border zones within healed myocardial infarctions, identifying critical structures to avoid such as coronary vessels. Taken together, OCT has the potential to decrease procedural time, decrease complications and increase chronic success rates of RFA therapy.

Widespread use of a real-time, compact, inexpensive imaging device may help to normalize the procedure, resulting in high success rates independent of institution or arrhythmia targeted.

168

Figure 6. 1. Clinical significance and utility of monitoring and guidance of ablation therapy using OCT.

169

6.3. Future Work

This work encourages the next step of translation, demonstrating in vivo monitoring and guidance of cardiac RFA therapy, improving OCT catheter probe designs, development of image analysis tools to extract optical properties, and provide correlated structure and function with an integrated optical mapping and OCT system.

6.3.1. In vivo monitoring and guidance of RFA

Further in vivo experiments are needed to assess the reproducibility of observing characteristic texture and morphological characteristics identified within ex vivo experiments in a living large animal. The first set of experiments will be used to evaluate if OCT imaging proves tissue electrode contact stability in comparison to fluoroscopy and impedance guidance. Secondly, in vivo imaging will allow validation of OCT imaging of critical structures by correlation with electro-anatomical mapping. Previous work has shown characteristic appearances of AV node[31, 88], Purkinje network[204], epicardial fat[191], untreated myocardium, coronaries within OCT imaging in ex vivo experiments in animals and recently in human hearts[90]. In vivo experiments will evaluate if these features are distinguishable in the presence of blood and motion, and using catheters with lower lateral resolutions than standard bench-top scanners. Lastly, in vivo imaging will allow for real time visualization of lesion formation. Since the prototype catheter is large in diameter, and the OCT catheter has to be strapped next to the RF catheter, initial demonstrations can be conducted in an open chest model. Second generation catheters which are smaller in diameter and integrated into the RF catheter will facilitate experiments in a closed chest procedure.

170

For demonstration of real time monitoring and guidance of a targeted arrhythmia, digitized time streams of power, temperature, electrograms, impedance, and EKG will allow for post procedural correlation of OCT image parameters with standard functional measurements. In addition to analysis of functional measurements and image analysis parameters derived from OCT image streams, number of lesions, mean duration of ablation and total time under fluoroscopic guidance can be compared for procedures guided with OCT guidance and without. This analysis addressed the key question of whether OCT guidance does in fact reduce procedure time, number of lesions delivered, and/or possibly incidence of over-treatment.

6.3.2. Catheter Development

The envisioned future use of OCT for intracardiac electrophysiology is an OCT catheter integrated into a RF catheter. An integrated OCT and RF catheter is preferred over a standalone OCT catheter because this will result in a small diameter probe, and provide a visual of the tissue directly where RF energy is being delivered. An integrated probe also has added advantage for adaptation, as the navigation will be the same as a standard RF catheter. Additionally, extra catheters, such as an echocardiography catheter are used during complex procedures. Therefore, for OCT to be used routinely, if there is a gain of function with imaging without the need to insert another catheter, it may increase the percentage of procedures in which OCT will be used. An integrated catheter is also needed for visualizing procedural complications. Reviewing all OCT images of steam pops or over treated lesions within ex vivo studies in Chapter 2, we can conclude that voids and cardiac perforations typically occurred in the center of the lesion.

Therefore, for OCT to be of clinical utility for visualizing early effects of damage, the

171

OCT catheter needs to be integrated into the RF catheter to enable visualizing dynamics directly under the RF electrode.

OCT catheters with various scanning geometries may be needed depending on the type of arrhythmia being targeted. For example, epicardial ablation may warrant a side viewing probe or a combination forward and side viewing catheter. Future generations of the forward imaging catheter probe will require a fiber with a stronger buffer, to increase the lifespan of the catheter. Since the catheter is designed entirely without metal, it can be used in conjunction with MRI procedures to provide high resolution imaging of areas of interest.

6.3.3. Polarization Sensitive Optical Coherence Tomography

A Polarization Sensitive Optical Coherence Tomography (PSOCT) system can be used in the future for artifact free imaging and quantitative birefringence measurements.

A PSOCT system can be implemented with a single spectrometer. Once such implementation using a superlumincent diode (SLD) centered at 1310 nm with a 70 nm

(FWHM) bandwidth as the light source, will record spectral interference fringes onto a

1024-pixel InGaAs line-scan camera (Goodrich) capable of more than 40,000 A-scans per second. A polarizer can be placed after the light source and an inline polarization controller will be used to optimize splitting of the light. Two in-line fiber-optic polarization controllers will be used to match polarization of sample and reference arms.

A polarization modulator (PM) will be placed in the sample arm. Using a method previously developed[205] the sample will be illuminated with three polarization states serially, recording an A-scan from each. The modulation rate of the PM can be synchronized to the readout rate of the spectrometer camera. From these data we can

172

reconstruct the reflectivity and the retardance and fast-axis of the birefringence of the tissue, corresponding to the conventional OCT image, the degree of myofiber organization, and possibly myofiber orientation, respectively. At an axial rate of 47kHz provided by the Goodrich line scan camera, with 3 measurements per axial scan, this reduces our effective axial scan rate to 15kHz. With 500 A lines per image, we can image at 30 frames per second, video rate. However, this will result in the lateral pixel dimension being 12.6µm. This is just above sampling our spot at the Nyquist rate.

Therefore, in the future we should investigate line scan cameras with higher axial scan rates or implementation of the PSOCT system with a swept source. Swept source lasers have been demonstrated using scanning filter implemented using grating and polygon mirrors[206] and fabry-perot tunable etalons[207, 208]. High tuning rates of 115kHz,

80nm tuning range, centered at 1325nm have been demonstrated with polygon filters[209]. Recently developed frequency swept laser technology, known as Fourier domain mode-locking (FDML) has been demonstrated to enable OCT imaging speeds

(370 kHz axial scan rates)[93, 210]. Higher axial scan rates will give greater flexibility in determining the frame rate and lines per frame. As demonstrated within our preliminary results imaging the myocardium in a swine in vivo, there were limited cases where the birefringence dependent band was observed while imaging at 20 fps. It was only observed within a subset of images of the right atria imaged from the Endocardium and the right ventricle imaged from the epicardium. We believe that with contraction, the orientation of the myofibers change and a change in the tissue birefringence. Therefore, imaging at a frame rate greater than 20 frames per second may be necessary to quantify birefringence and observe dynamic changes in birefringence due to RF energy delivery.

173

6.3.4. Correlation of attenuation coefficient with lesion depth

Radiofrequency ablation causes thermal damage due to resistive heating, producing an area of coagulation necrosis[6]. Thermal damage of the myocardium has been shown to cause changes in the optical properties of tissue, in particular anisotropy coefficient, scattering coefficient[162-164], and birefringence[174]. Previous work has shown that scattering coefficient changes linearly with increasing temperature.

Polarization sensitive OCT or an OCT system with polarization diverse detection will allow for artifact free imaging, and allow quantification of additional analysis parameters: attenuation and backscattering coefficients. Previous work has shown that tissue temperature is highly correlated with lesion depth. However, the temperature sensor on the RF catheter reflects the temperature of the catheter and not within the tissue. With quantitative analysis of dynamic changes in attenuation coefficient, we can address the question if OCT can provide surrogate for lesion depth.

6.3.5. Technology development for cardiac tissue characterization using OCT

Within traditional OCT scanners, achromatic lenses are used to focus light in the tissue. With the use of Gaussian optics, there is an inherent tradeoff between the lateral resolution and depth of focus. Therefore, if one wants to achieve a higher lateral resolution, there is a decrease in image penetration. Within our work imaging and quantifying myofiber orientation within various animal models, there were two important observations. First, the ease of myofiber visibility within planes parallel to the sample surface varied by species. Myofibers were visible within planes parallel to the endocardial surface within freshly excised swine hearts, without the need for fixation or

174

optical clearing imaging with a 1310nm OCT system with 10µm axial and lateral resolutions. Myofibers were visible within planes parallel to the endocardial surface of rabbit hearts with fixation and optical clearing imaging with a 1310nm OCT system with

10µm axial and lateral resolutions. However, myofibers were not visible on a reproducible basis within mouse hearts with fixation and/or clearing imaging with a

1310nm OCT system with 10µm axial and lateral resolutions and visible with fixation within the first 50 um with an 830nm FDOCT with 4µm axial and 7µm lateral resolutions. Secondly, within swine and rabbit experimental data, myofiber visibility was limited to the first 500um below the sample surface. This opens the opportunity for technology development for high resolution imaging, with a system that maintains a high lateral resolution as a function of depth. In addition, development of optical clearing protocols to increase the depth to which myofibers can be reproducibly visible within a range of animal models.

Both high axial and high lateral resolutions are necessary for image myofibers in three dimensions. High axial resolution is achieved through the use of light sources with broad bandwidths. In order to achieve ultrahigh axial resolution (< 5 µm in air) several broadband light sources have been employed, including broadband solid state lasers[36], supercontinuum generation in highly nonlinear fiber[211] and multiplexed SLEDs[212].

Water absorption and dispersion are other important factors to consider when designing

OCT systems with ultra high axial resolution. Traditional achromatic lens may not be helpful to achieve a goal of marinating a high lateral resolution over a large imaging range. One such method for increasing the imaging range without compromising the lateral resolution is through the use of an axicon lens[213].

175

The optical clearing agents glycerol and Dimethylsulfoxide (DMSO) have been shown in literature to be to be effective in increasing image penetration within tissue[214,

215]. Within preliminary studies, we observed an increased image penetration as a result of treating the tissue with DMSO. However, myofiber visibility was greatly enhanced with the application of glycerol. Ideally, an optical clearing agent is needed that increases image penetration and myofiber visibility. Further analysis on the effect of optical clearing agents on the absorption coefficient, scattering coefficient, and anisotropy coefficient will help to optimize protocols for reproducibly observing myofibers within animal models of heart disease.

In addition to providing morphological information, there is potential to use OCT to provide correlated structural and functional information. This can be done by combining an OCT system with an Optical mapping system. Optical mapping is a non- contact imaging technique that uses fast voltage sensitive dyes to create 2D images of conduction patterns on cardiac tissue preparations with high spatial and temporal resolution. However, optical mapping provides functional data without correlated 3D structural data. An integrated optical mapping and OCT scanner will provide correlated

2D functional and 3D structural images of cardiac tissue. The OCT scanner described in chapter 3 was designed and developed to enable such studies. A dichroic mirror can be used to separate the infrared OCT signal and the visible fluorescence signal. A custom mount was made to fix six green LED under the OCT scanner, to serve as an illumination source for optical mapping. The scanner has a long working distance and large field of view. The current system provides 18um lateral resolution, and with future generations, a system that provides 10um or less lateral resolution will enable imaging of fiber

176

orientation in addition to obtaining optical mapping data. In addition to correlated fiber orientation and conduction velocity measurements, the integrated system can be used to study the impact of structural changes, such as radiofrequency ablation lesions, infarction, fibrosis, etc, on conduction patterns. With the integration of an OCT scanner and a fluorescence conduction mapping system, both the structure and function of cardiac tissue can be obtained simultaneously. With this new information, the structural changes in the cardiac tissue that result in abnormal conduction of action potentials can be readily identified and evaluated for further study. In addition, structure provided by

OCT images can be inputs for boundary conditions to solve the optical mapping inverse problem.

177

Bibliography

1. W. Rosamond, K. Flegal, K. Furie, A. Go, K. Greenlund, N. Haase, S. M.

Hailpern, M. Ho, V. Howard, B. Kissela, S. Kittner, D. Lloyd-Jones, M.

McDermott, J. Meigs, C. Moy, G. Nicho, C. O’Donnell, V. Roger, P. Sorlie, J.

Steinberger, T. Thom, M. Wilson, and Y. Hong, "Heart Disease and Stroke

Statistics2008 Update: A Report From the American Heart Association Statistics

Committee and Stroke Statistics Subcommittee," Circulation 117, e25-e146

(2008).

2. V. Obelienius, A. Knepa, R. Ambartzumian, E. Markin, and E. Koshelev,

"Transvenous ablation of the atrioventricular conduction system by laser

irradiation under endoscopic control.," Lasers in Medical Science 5, 469-474

(1985).

3. J. J. Gallagher, R. H. Svenson, J. H. Kasell, L. D. German, G. H. Bardy, A.

Broughton, and G. Critelli, "Catheter technique for closed-chest ablation of the

atrioventricular conduction system.," New England Journal of Medicine 306, 194-

200 (1982).

4. M. M. Scheinman, F. Morady, D. S. Hess, and R. Gonzalez, "Catheter-induced

ablation of the atrioventricular junction to control refractory supraventricular

arrhythmias.," The journal of the American Medical Association 248, 851-855

(1982).

5. S. K. Huang, S. Bharati, A. R. Graham, M. Lev, F. I. Marcus, and R. C. Odell,

"Closed chest catheter desiccation of the atrioventricular junction using

178

radiofrequency energy--a new method of catheter ablation.," Journal of American

College of Cardiology 9, 349-358 (1987).

6. S. K. S. Huang and M. A. Wood, Catheter Ablation of Cardiac Arrhythmias

(Saunders, 2006).

7. J. Alaeddini, M. A. Wood, B. P. Lee, and K. A. Ellenbogen, "Incidence, Time

Course, and Characteristics of Microbubble Formation During Radiofrequency

Ablation of Pulmonary Veins with an 8-mm Ablation Catheter," Pacing Clin

Electrophysiol 29, 979-984 (2006).

8. G. Stabile, E. Bertaglia, G. Senatore, A. de Simone, F. Zerbo, G. Carreras, P.

Turco, P. Pascotto, and M. Fazzari, "Feasibility of pulmonary vein ostia

radiofrequency ablation in patients with atrial fibrillation: a multicenter study

(CACAF pilot study)," Pacing Clin Electrophysiol. 26 (1 pt 2), 284-287 (2003).

9. N. Marrouche, J. Guenther, N. Segerson, M. Daccarett, H. Rittger, H. Marschang,

V. Schibgilla, M. Schmidt, G. Ritscher, G. Noelker, and J. Brachmann,

"Randomized comparison between open irrigation technology and intracardiac-

echo-guided energy delivery for pulmonary vein antrum isolation: procedural

parameters, outcomes, and the effect on esophageal injury.," J Cardiovasc

Electrophysiol. 18, 583-588 (2007).

10. R. Brockman, "Cardiac Ablation Catheters Generic Arrhythmia Indications for

Use; Guidance for Industry," (FDA Center for Devices and Radiological Health.

Cardiac Electrophysiology and Monitoring Branch Division of Cardiovascular

and Respiratory Devices Office of Device Evaluation, Rockville, MD, 2002).

179

11. R. Cappato, H. Calkins, S.-A. Chen, W. Davies, Y. Iesaka, J. Kalman, Y.-H. Kim,

G. Klein, D. Packer, and A. Skanes, "Worldwide Survey on the Methods,

Efficacy, and Safety of Catheter Ablation for Human Atrial Fibrillation,"

Circulation 111, 1100-1105 (2005).

12. T. Dickfeld, R. Kato, M. Zviman, S. Nazarian, H. Ashikaga, A. C. Lardo, R. D.

Berger, H. Calkins, and H. Halperin, "Characterization of Acute and Subacute

Radiofrequency Ablation Lesions with Non-enhanced Magnetic Resonance

Imaging," Heart Rhythm 4, 208-214 (2007).

13. T. Dickfeld, R. Kato, M. Zviman, S. Lai, G. Meininger, A. C. Lardo, A. Roguin,

D. Blumke, R. Berger, H. Calkins, and H. Halperin, "Characterization of

Radiofrequency Ablation Lesions with Gadolinium-Enhanced Cardiovascular

Magnetic Resonance Imaging," J Am Coll Cardiol 47, 370-378 (2006).

14. S. Abbara, J. C. Desai, R. C. Cury, J. Butler, K. Nieman, and V. Reddy, "Mapping

epicardial fat with multi-detector computed tomography to facilitate percutaneous

transepicardial arrhythmia ablation.," European Journal of Radiology 57, 417-422

(2006).

15. S. Sarin, C. Wenger, A. Marwaha, A. Qureshi, B. D. Go, C. A. Woomert, K.

Clark, L. A. Nassef, and J. Shirani, "Clinical significance of epicardial fat

measured using cardiac multislice computed tomography.," American Journal of

Cardiology 102, 767-771 (2008).

16. M. David Schwartzman, R. John Nosbisch, and R. Debra Housel,

"Echocardiographically guided left atrial ablation: Characterization of a new

technique," Heart Rhythm Society 3, 930-938 (2006).

180

17. T. Dickfeld, R. Kato, M. Zviman, S. Lai, G. Meininger, A. C. Lardo, A. Roguin,

D. Blumke, R. Berger, H. Calkins, and H. Halperin, "Characterization of

Radiofrequency Ablation Lesions With Gadolinium-Enhanced Cardiovascular

Magnetic Resonance Imaging," Journal of the American College of Cardiology

47, 370–378 (2008).

18. D. S. Khoury, L. Rao, C. Ding, H. Sun, K. A. Youker, D. Panescu, and S. F.

Nagueh, "Localizng and Quantifying Ablation Lesions in the Left Ventricle by

Myocardial Contrast Echocardiography," J Cardiovasc Electrophysiol 15, 1078-

1087 (2004).

19. R. Citro, V. Ducceschi, A. Salustri, M. Santoro, M. Salierno, and G. Gregorio,

"Intracardiac echocardiography to guide transseptal catheterization for

radiofrequency catheter ablation of left-sided accessory pathways: two case

reports," Cardiovascular Ultrasound 2(2004).

20. J.-F. Ren and F. E. Marchinski, "Utility of Intracardiac Echocardiography in Left

Heart Ablation for Tachyarrhythmias," Echocardiography 24, 533-540 (2007).

21. P. Kovoor, M. Daly, J. Pouliopoulos, V. Eipper, B. Dewsnap, and D. Ross,

"Comparison of unipolar versus bipolar ablation and single electrode control

versus simultaneous multielectrode temperature control.," J Interv Card

Electrophysiol 19, 85-93 (2007).

22. E. Gerstenfeld, J. Jacobson, V. Bazan, S. Lazar, J. Tomaszewski, F. Marchlinski,

and J. Michele, "Comparison of high power, medium power, and irrigated-tip

ablation strategies for pulmonary vein isolation in a canine model.," J Cardiovasc

Electrophysiol 18, 849-853 (2007).

181

23. S. Nazarian, S. V. Kantsevoy, M. M. Zviman, F. A. r. Matsen, H. Calkins, R. D.

Berger, and H. R. Halperin, "Feasibility of endoscopic guidance for nonsurgical

transthoracic atrial and ventricular epicardial ablation.," Heart Rhythm 5, 1115-

1119 (2008).

24. V. Y. Reddy, P. Neuzil, Z. J. Malchano, R. Vijaykumar, R. Cury, S. Abbara, J.

Weichet, C. D. McPherson, and J. N. Ruskin, "View-synchronized robotic image-

guided therapy for atrial fibrillation ablation: experimental validation and clinical

feasibility.," Circulation 115, 2705-2714 (2007).

25. D. D. Streeter, H. M. Spotnitz, D. J. Patel, J. J. Ross, and E. H. Sonnenblick,

"Fiber Orientation in the Canine Left Ventricle during Diastole and Systole,"

Circulation Research 24, 339 (1969).

26. D. D. Streeter, R. N. Vaishnav, D. J. Patel, H. M. Spotnitz, J. J. Ross, and E. H.

Sonnenblick, "Stress Distribution in the Canine Left Ventricle during Diastole and

Systole," Biophys J 10, 345-363 (1970).

27. L. Zhukov and A. H. Barr, "Oriented Tensor Reconstruction: Tracing Neural

Pathways from Diffusion Tensor MRI," IEEE Visualization 2002, Proceedings of

Vis 2002, 387-394 (2002).

28. L. Zhukov and A. H. Barr, "Heart-Muscle Fiber Reconstruction from Diffusion

Tensor MRI," Proceedings of the 14th IEEE Visualization 2003, 79 (2003).

29. J. Chen, S.-K. Song, W. Liu, M. McLean, J. S. Allen, J. Tan, S. A. Wickline, and

X. Yu, "Remodeling of cardiac fiber structure after infarction in rats quantified

with diffusion tensor MRI," Am J Physiol Heart Circ Physiol 285, H946-H954

(2003).

182

30. A. Kanai and G. Salama, "Optical Mapping Reveals that Repolarization Spreads

Anistropically and is Guided by Fiber Orientation in Guinea Pig Hearts,"

Circulation Research 77, 784-802 (1995).

31. W. Hucker, C. Ripplinger, C. P. Fleming, V. Fedorov, A. M. Rollins, and I. R.

Efimov, "Bimodal Biophotonic Imaging of the Structure-Function Relationship in

Cardiac Tissue," J Biomed Opt 13, 054012 (2008).

32. F. J. Vetter, S. B. Simons, S. Mironov, C. J. Hyatt, and A. M. Pertsov, "Epicardial

Fiber Organization in Swine Right Ventricle and Its Impact on Propagation,"

Circulation Research 96, 244 (2004).

33. I. R. Efimov, V. P. Nikolski, and G. Salama, "Optical Imaging of the heart,"

Circulation Research 95, 21-33 (2004).

34. D. Huang, E. A. Swanson, C. P. Lin, J. S. Schuman, W. G. Stinson, W. Chang, M.

R. Hee, T. Flotte, K. Gregor, C. A. Puliafito, and J. G. Fujimoto, "Optical

Coherence Tomography," Science 254, 1178-1181 (1991).

35. J. A. Izatt, M. D. Kulkarni, W. Hsing-Wen, K. Kobayashi, and M. V. Sivak, Jr,

"Optical coherence tomography and microscopy in gastrointestinal tissues,"

Selected Topics in Quantum Electronics, IEEE Journal of 2, 1017-1028 (1996).

36. W. Drexler, U. Morgner, F. X. Kartner, C. Pitris, S. A. Boppart, X. D. Li, E. P.

Ippen, and J. G. Fujimoto, "In Vivo Ultrahigh-Resolution Optical Coherence

Tomography," Optics Letters 24, 1221-1223 (1999).

37. J. M. Schmitt, A. Knuttel, and R. F. Bonner, "Measurment of optical properties of

biological tissues by low-coherence reflectometry," Applied Optics 32, 6032-

6041 (1993).

183

38. A. F. Fercher, C. K. Hitzenberger, G. Kamp, and S. Y. El-Zaiat, "Measurement of

intraocular distances by backscattering spectral interferometry," Optics

Communications 117, 43-48 (1995).

39. M. Wojtkowski, R. Leitgeb, A. Kowalczyk, T. Bajraszewski, and A. F. Fercher,

"In vivo human retinal imaging by Fourier domain optical coherence

tomography," Journal of Biomedical Optics 7, 457-463 (2002).

40. M. Choma, M. Sarunic, C. Yang, and J. Izatt, "Sensitivity advantage of swept

source and Fourier domain optical coherence tomography," Opt. Express 11,

2183-2189 (2003).

41. R. Leitgeb, C. Hitzenberger, and A. Fercher, "Performance of fourier domain vs.

time domain optical coherence tomography," Opt. Express 11, 889-894 (2003).

42. M. Wojtkowski, A. Kowalczyk, R. Leitgeb, and A. F. Fercher, "Full range

complex spectral optical coherence tomography technique in eye imaging," Opt.

Lett. 27, 1415-1417 (2002).

43. R. A. Leitgeb, C. K. Hitzenberger, A. F. Fercher, and T. Bajraszewski, "Phase-

shifting algorithm to achieve high-speed long-depth-range probing by frequency-

domain optical coherence tomography," Opt. Lett. 28, 2201-2203 (2003).

44. E. Götzinger, M. Pircher, R. Leitgeb, and C. Hitzenberger, "High speed full range

complex spectral domain optical coherence tomography," Opt. Express 13, 583-

594 (2005).

45. Y. K. Tao, M. Zhao, and J. A. Izatt, "High-speed complex conjugate resolved

retinal spectral domain optical coherence tomography using sinusoidal phase

modulation," Opt. Lett. 32, 2918-2920 (2007).

184

46. M. Sarunic, M. A. Choma, C. Yang, and J. A. Izatt, "Instantaneous complex

conjugate resolved spectral domain and swept-source OCT using 3x3 fiber

couplers," Opt. Express 13, 957-967 (2005).

47. M. V. Sarunic, B. E. Applegate, and J. A. Izatt, "Real-time quadrature projection

complex conjugate resolved Fourier domain optical coherence tomography.,"

Optics Letters 31, 2426-2428 (2006).

48. T. Bajraszewski, M. Wojtkowski, M. Szkulmowski, A. Szkulmowska, R. Huber,

and A. Kowalczyk, "Improved spectral optical coherence tomography using

optical frequency comb," Opt. Express 16, 4163-4176 (2008).

49. Z. Hu, Y. Pan, and A. M. Rollins, "Analytical model of spectrometer-based two-

beam spectral interferometry," Appl. Opt. 46, 8499-8505 (2007).

50. Z. Hu and A. Rollins, "Fourier domain optical coherence tomography with a

linear-in-wavenumber spectrometer.," Opt Lett 32, 3524-3527 (2007).

51. J. F. de Boer and T. E. Milner, "Review of polarization sensitive optical

coherence tomography and Stokes vector determination," J. Biomed. Opt. 7, 359-

371 (2002).

52. S. Nadkarni, M. Pierce, B. Park, J. de Boer, P. Whittaker, B. Bouma, J. Bressner,

E. Halpern, S. Houser, and G. Tearney, "Measurement of collagen and smooth

muscle cell content in atherosclerotic plaques using polarization-sensitive optical

coherence tomography.," J Am Coll Cardiol. 49, 1474-1481 (2007).

53. S. Nadkarni, B. Bouma, J. de Boer, and G. Tearney, "Evaluation of collagen in

atherosclerotic plaques: the use of two coherent laser-based imaging methods,"

Lasers Med Sci (2008).

185

54. W.-C. Kuo, M.-W. Hsiung, J.-J. Shyu, N.-K. Chou, and P.-N. Yang, "Assessment

of arterial characteristics in human atherosclerosis by extracting optical properties

from polarization-sensitive optical coherence tomography," Optics Express 16,

8117-8125 (2008).

55. C.-W. Sun, Y.-M. Wang, L.-S. Lu, C.-W. Lu, I.-J. Hsu, M.-T. Tsai, C. C. Yang,

Y.-W. Kiang, and C.-C. Wu, "Myocardial tissue characterization based on a

polarization-sensitve optical coherence tomography system with an ultrashort

pulsed laser," J Biomed Opt 11, 054016 (2006).

56. J. F. de Boer, S. M. Srinivas, A. Malekafzali, Z. Chen, and J. S. Nelson, "Imaging

thermally damaged tissue by polarization sensitive optical coherence

tomography," Optics Express 3, 212-218 (1998).

57. K. Schoenenberger, J. Bill W. Colston, D. J. Maitland, L. B. D. Silva, and M. J.

Everett, "Mapping of birefringence and thermal damage in tissue by use of

polarization-sensitive optical coherence tomography," Applied Optics 37, 6026-

6036 (1998).

58. E. Götzinger, M. Pircher, W. Geitzenauer, C. Ahlers, B. Baumann, S. Michels, U.

Schmidt-Erfurth, and C. K. Hitzenberger, "Retinal pigment epithelium

segmentation by polarization sensitive optical coherence tomography," Optics

Express 16, 16410-16422 (2008).

59. B. Cense, T. C. Chen, B. H. Park, M. D. Pierce, and J. F. d. Boer, "In vivo depth-

resolved birefringence measurements of the human retinal nerve fiber layer by

polarization-sensitive optical coherence tomography," optics letters 27, 1610

(2002).

186

60. M. Yamanari, M. Miura, S. Makita, T. Yatagai, and Y. Yasuno, "Phase

retardation measurement of retinal nerve fiber layer by polarization-sensitive

spectral-domain optical coherence tomography and scanning laser polarimetry.,"

Journal of Biomedical Optics 13, 014013 (2008).

61. E. Götzinger, M. Pircher, and C. K. Hitzenberger, "High speed spectral domain

polarization sensitive optical coherence tomography of the human retina.," Optics

Express 13, 10217-10229 (2005).

62. W. Oh, S. Yun, B. Vakoc, M. Shishkov, A. Desjardins, B. Park, J. de Boer, G.

Tearney, and B. Bouma, "High-speed polarization sensitive optical frequency

domain imaging with frequency multiplexing.," Optics Express 16, 1096-1103

(2008).

63. N. R. Munce, A. Mariampillai, B. A. Standish, M. Pop, K. J. Anderson, G. Y. Liu,

T. Luk, B. K. Courtney, G. A. Wright, I. A. Vitkin, and V. X. D. Yang,

"Electrostatic forward-viewing scanning probe for doppler optical coherence

tomography using a dissipative polymer catheter," Opt Lett 33, 657-659 (2008).

64. J. Wu, M. Conry, C. Gu, F. Wang, Z. Yaqoob, and C. Yang, "Paired-angle-

rotation scanning optical coherence tomography forward-imaging probe," Opt

Lett 31, 1265-1267 (2006).

65. X. Liu, M. J. Cobb, and Y. Chen, "Rapid-scanning forward-imaging miniature

endoscope for real-time optical coherence tomography," Opt Lett 29, 1763-1765

(2004).

66. X. D. Li, X. M. Liu, Y. C. Chen, M. J. Cobb, and M. B. Kimmey, "Development

of a Fast Scanning Miniature Probe and Methods of Dispersion Management for

187

High-resolution Optical Coherence Tomography," in IEEE EMBS, 2004), 5289-

5291.

67. P. R. Herz, Y. Chen, A. D. Aguirre, K. Schneider, P. Hsiung, and J. G. Fujimoto,

"Micromotor endoscope catheter for in vivo, ultrahigh resolution optical

coherence tomography," Opt Lett 29, 2261-2263 (2004).

68. Y. Wang, M. Bachman, G.-P. Li, S. Guo, B. J. F. Wong, and Z. Chen, "Low-

voltage polymer-based scanning cantilever for in vivo optical coherence

tomography," Opt Lett 30, 53-55 (2005).

69. D. Yelin, B. E. Bouma, and G. J. Tearney, "Volumetric sub-surface imaging using

spectrally encoded endoscopy," Optics Express 16, 1749-1757 (2008).

70. K. H. Kim, B. H. Park, G. N. Maguluri, T. W. Lee, F. J. Rogomentich, M. G.

Bancu, B. E. Bouma, J. F. de Boer, and J. J. Bernstein, "Two-axis magnetically-

driven MEMS scanning catheter for endoscopic high-speed optical coherence

tomography," Optics Express 15, 18130-18140 (2007).

71. P. Tran, D. Mukai, M. Brenner, and Z. Chen, "In vivo endoscopic optical

coherence tomography by use of a rotational microelectromechanical system

probe.," Optics Express 29, 1236-1238 (2004).

72. J. Xi, L. Huo, Y. Wu, M. J. Cobb, J. H. Hwang, and X. Li, "High-resolution OCT

balloon imaging catheter with astigmatism correction," Optics Letters 34, 1943

(2009).

73. A. P. Burke, A. Farb, G. T. Malcom, Y.-h. Liang, J. Smialek, and R. Virmani,

"Coronary Risk Factors and Plaque Morphology in Men with Coronary Disease

Who Died Suddenly," N Engl J Med 336, 1276-1282 (1997).

188

74. K. Teruyoshi, A. Takashi, K. Takahiro, O. Hiroyuki, W. Nozomi, T. Eiji, N. Yoji,

S. Renan, S. Yoshito, and Y. Kiyoshi, "Measurement of the thickness of the

fibrous cap by optical coherence tomography," American heart journal 152,

755.e751-755.e754 (2006).

75. H. Yabushita, B. E. Bouma, S. L. Houser, H. T. Aretz, I. K. Jang, K. H.

Schlendorf, C. R. Kauffman, M. Shishkov, D. H. Kang, E. F. Halpern, and G. J.

Tearney, "Characterization of human atherosclerosis by optical coherence

tomography," Circulation 106, 1640-1645 (2002).

76. M. Cilingiroglu, J. H. Oh, B. Sugunan, N. J. Kemp, J. Kim, S. Lee, H. N. Zaatari,

D. Escobedo, S. Thomsen, T. E. Milner, and M. D. Feldman, "Detection of

vulnerable plaque in a murine model of atherosclerosis with optical coherence

tomography.," Catheter Cardiovasc Interv. 67, 915-923 (2006).

77. v. d. Meer, "Quantitative optical coherence tomography of arterial wall

components," Lasers in Medical Science 20, 45 (2005).

78. v. d. Meer, "Localized measurement of optical attenuation coefficients of

atherosclerotic plaque constituents by quantitative optical coherence

tomography," IEEE Transactions on Medical Imaging 24, 1369 (2005).

79. G. J. Tearney, H. Yabushita, S. L. Houser, H. T. Aretz, I.-K. Jang, K. H.

Schlendorf, C. R. Kauffman, M. Shishkov, E. F. Halpern, and B. E. Bouma,

"Quantification of Macrophage Content in Atherosclerotic Plaques by Optical

Coherence Tomography," Circulation 107, 113-119 (2003).

80. O. C. Raffel, G. J. Tearney, D. D. Gauthier, E. F. Halpern, B. E. Bouma, and I.-K.

Jang, "Relationship Between a Systemic Inflammatory Marker, Plaque

189

Inflammation, and Plaque Characteristics Determined by Intravascular Optical

Coherence Tomography," Arterioscler Thromb Vasc Biol 27, 1820-1827 (2007).

81. J. Oh, M. D. Feldman, J. Kim, P. Sanghi, D. Do, J. J. Mancuso, N. Kemp, M.

Cilingiroglu, and T. E. Milner, "Detection of macrophages in atherosclerotic

tissue using magnetic nanoparticles and differential phase optical coherence

tomography.," Journal of Biomedical Optics 13, 054006 (2008).

82. P. Moore, P. Barlis, J. Spiro, G. Ghimire, M. Roughton, C. Di Mario, W. Wallis,

C. Ilsley, A. Mitchell, M. Mason, R. Kharbanda, P. Vincent, S. Sherwin, and M.

Dalby, "A randomized optical coherence tomography study of coronary stent strut

coverage and luminal protrusion with rapamycin-eluting stents.," JACC

Cardiovasc Interv. 2, 437-444 (2009).

83. N. Gonzalo, H. M. Garcia-Garcia, P. W. Serruys, K. H. Commissaris, H. Bezerra,

P. Gobbens, M. Costa, and E. Regar, "Reproducibility of quantitative optical

coherence tomography for stent analysis.," EuroIntervention 5, 224-232 (2009).

84. G. T. Bonnema, K. O. Cardinal, S. K. Williams, and J. K. Barton, "An automatic

algorithm for detecting stent endothelialization from volumetric optical coherence

tomography datasets.," Physics in Medicine and Biology 53, 3083-3098 (2008).

85. H. Otake, J. Shite, J. Ako, T. Shinke, Y. Tanino, D. Ogasawara, T. Sawada, N.

Miyoshi, H. Kato, B. K. Koo, Y. Honda, P. J. Fitzgerald, and K. Hirata, "Local

determinants of thrombus formation following sirolimus-eluting stent

implantation assessed by optical coherence tomography.," JACC Cardiovasc

Interv. 2, 459-466 (2009).

190

86. G. J. Tearney, S. Waxman, M. Shishkov, B. J. Vakoc, M. J. Suter, M. I. Freilich,

A. E. Desjardins, W.-Y. Oh, L. A. Bartlett, M. Rosenberg, and B. E. Bouma,

"Three-Dimensional Coronary Artery Microscopy by Intracoronary Optical

Frequency Domain Imaging," JACC: Cardiovascular Imaging 1, 752 (2008).

87. M. W. Jenkins, R. S. Wade, Y. Cheng, A. M. Rollins, and I. R. Efimov, "Optical

Coherence Tomography Imaging of the Purkinje Network," J Cardiovasc

Electrophysiol (2005).

88. M. Gupta, A. M. Rollins, J. A. Izatt, and I. R. Efimov, "Imaging of the

atrioventricular node using optical coherence tomography," J Cardiovasc

Electrophysiol 13, 95 (2002).

89. C. P. Fleming, C. Ripplinger, B. Webb, I. R. Efimov, and A. M. Rollins,

"Quantification of Cardiac Fiber Orientation Using Optical Coherence

Tomography," J Biomed Opt 13, 030505 (2008).

90. C. M. Ambrosi, N. Moazami, A. M. Rollins, and I. R. Efiomv, "Virtual histology

of the human heart using optical coherence tomography," Journal of Biomedical

Optics 14, 054002 (2009).

91. J. Villard, M. Feldman, J. Kim, T. Milner, and G. Freeman, "Use of a blood

substitute to determine instantaneous murine right ventricular thickening with

optical coherence tomography.," Circulation 105, 1843-1849 (2002).

92. M. W. Jenkins, P. Patel, H. Deng, M. M. Montano, M. Watanabe, and A. M.

Rollins, "Phenotyping transgenic embryonic murine hearts using optical

coherence tomography.," Applied Optics 46, 1776-1781 (2007).

191

93. M. W. Jenkins, D. C. Adler, M. Gargesha, R. Huber, F. Rothenberg, J. Belding2,

M. Watanabe, D. L. Wilson, J. G. Fujimoto, and A. M. Rollins, "Ultrahigh-speed

optical coherence tomography imaging and visualization of the embryonic avian

heart using a buffered Fourier Domain Mode Locked laser," Optics Express 15,

6251-6267 (2007).

94. M. W. Jenkins, O. Q. Chughtai, A. N. Basavanhally, M. Watanabe, and A. M.

Rollins, "In vivo gated 4D imaging of the embryonic heart using optical

coherence tomography," Journal of Biomedical Optics 12, 030505 (2007).

95. M. Gargesha, M. W. Jenkins, D. L. Wilson, and A. M. Rollins, "High temporal

resolution OCT using image-based retrospective gating.," Optics Express 17,

10786-10799 (2009).

96. J. Männer, L. Thrane, K. Norozi, and T. M. Yelbuz, "High-resolution in vivo

imaging of the cross-sectional deformations of contracting embryonic heart loops

using optical coherence tomography.," Developmental Dynamics 237, 953-961

(2008).

97. S. Yazdanfar, M. D. Kulkarni, and J. A. Izatt, "High Resolution Imaging of In

Vivo Cardiac Dynamics Using Color Doppler Optical Coherence Tomography,"

Optics Express 1, 424-431 (1997).

98. V. X. D. Yang, M. L. Gordon, B. Qi, J. Pekar, S. Lo, E. Seng-Yue, A. Mok, B. C.

Wilson, and I. A. Vitkin, "High speed, wide velocity dynamic range Doppler

optical coherence tomography (Part II): Imaging in vivo cardiac dynamics of

Xenopus laevis," Optics Express 11, 1650-1658 (2003).

192

99. A. Davis, J. Izatt, and F. Rothenberg, "Quantitative Measurement of Blood Flow

Dynamics in Embryonic Vasculature Using Spectral Doppler Velocimetry," The

Anatomical Record 292, 311-319 (2009).

100. I. V. Larina, N. Sudheendran, M. Ghosn, J. Jiang, A. Cable, K. V. Larin, and M.

Dickinson, "Live imaging of blood flow in mammalian embryos using Doppler

swept-source optical coherence tomography.," Journal of Biomedical Optics 13,

060506 (2008).

101. S. Rugonyi, C. Shaut, A. Liu, K. Thornburg, and R. K. Wang, "Changes in wall

motion and blood flow in the outflow tract of chick embryonic hearts observed

with optical coherence tomography after outflow tract banding and vitelline-vein

ligation.," Physics in Medicine and Biology 53, 5077-5091 (2008).

102. I. V. Larina, S. Ivers, S. Syed, M. E. Dickinson, and K. V. Larin, "Hemodynamic

measurements from individual blood cells in early mammalian embryos with

Doppler swept source OCT.," Optics Letters 34, 986-988 (2009).

103. M. Liebling, A. S. Forouhar, M. Gharib, S. E. Fraser, and M. E. Dickinson, "Four-

dimensional cardiac imaging in living embryos via postacquisition

synchronization of nongated slice sequences," Journal of Biomedical Optics 10,

054001-054010 (2005).

104. A. E. Desjardins, B. J. Vakoc, W. Y. Oh, S. M. Motaghiannezam, G. J. Tearney,

and B. E. Bouma, "Angle-resolved optical coherence tomography with sequential

angular selectivity for speckle reduction.," Optics Express 15, 6200-6209 (2007).

193

105. A. E. Desjardins, B. J. Vakoc, G. J. Tearney, and B. E. Bouma, "Backscattering

spectroscopic contrast with angle-resolved optical coherence tomography.,"

Optics Letters 32, 3158-3160 (2007).

106. H. Wang and A. M. Rollins, "Speckle reduction in optical coherence tomography

using angular compounding by B-scan Doppler-shift encoding.," Journal of

Biomedical Optics 14, 030512 (2009).

107. M. Pircher, E. Gotzinger, R. Leitgeb, A. F. Fercher, and C. K. Hitzenberger,

"Speckle reduction in optical coherence tomography by frequency

compounding.," Journal of Biomedical Optics 8, 565-569 (2003).

108. J. M. Schmitt, S. H. Xiang, and K. M. Yung, "Speckle in Optical Coherence

Tomography," Journal of Biomedical Optics 4, 95-105 (1999).

109. J. Rogowska and M. E. Brezinski, "Image processing techniques for noise

removal, enhancement and segmentation of cartilage OCT images," Physics in

Medicine and Biology 47, 641-655 (2002).

110. X. Qi, Y. Pan, Z. Hu, W. Kang, J. E. Willis, K. Olowe, M. V. S. Jr., and A. M.

Rollins, "Automated quantification of colonic crypt morphology using integrated

microscopy and optical coherence tomography," Journal of Biomedical Optics 13,

054055 (2008).

111. J. Rogowska and M. E. Brezinski, "Evaluation of the Adaptive Speckle

Suppression Filter for Coronary Optical Coherence Tomography Imaging," IEEE

Transactions on Image Processing 19, 1261-1266 (2000).

194

112. D. C. Adler, T. H. Ko, and J. G. Fujimoto, "Speckle reduction in optical

coherence tomography images byvuse of a spatially adaptive wavelet filter,"

Optics Letters 29, 2878-2880 (2004).

113. P. Puvanathasan and K. Bizheva, "Speckle noise reduction algorithm for optical

coherence tomography based on interval type II fuzzy set," Optics Express 15,

15747-15758 (2007).

114. A. Ozcan, A. Bilenca, A. E. Desjardins, B. E. Bouma, and G. J. Tearney, "Speckle

reduction in optical coherence tomography images using digital filtering," J Opt

Soc Am A Opt Image Sci Vis 24, 1901-1910 (2007).

115. S. Bhat, I. V. Larina, K. V. Larin, M. E. Dickinson, and M. Liebling1, "Multiple-

cardiac-cycle noise reduction in dynamic optical coherence tomography of the

embryonic heart and vasculature," Optics Letters 34, 3704-3706 (2009).

116. M. Gargesha, M. W. Jenkins, A. M. Rollins, and D. L. Wilson, "Denoising and

4D visualization of OCT images," Optics Express 16, 12313-12333 (2008).

117. J. A. Izatt, M. D. Kulkarni, S. Yazdanfar, J. K. Barton, and A. J. Welch, "In Vivo

Bidirectional Color Doppler Flow Imaging of Picoliter Blood Volumes Using

Optical Coherence Tomography," Optics Letters 22, 1439-1441 (1997).

118. J. Schmitt, "Restoration of Optical Coherence Images of Living Tissue Using the

CLEAN Algorithm," Journal of Biomedical Optics 3, 66-75 (1998).

119. D. Piao, Q. Zhu, N. K. Dutta, S. Yan, and L. L. Otis, "Cancellation of coherent

artifacts in optical coherence tomography imaging," Applied Optics 40, 5124-

5131 (2001).

195

120. T. S. Ralston, D. L. Marks, F. Kamalabadi, and S. A. Boppart, "Deconvolution

Methods for Mitigation of Transverse Blurring in Optical Coherence

Tomography," IEEE Transactions on Image Processing 14, 1254-1264 (2005).

121. Y. Yasuno, J.-i. Sugisaka, Y. Sando, Y. Nakamura, S. Makita, M. Itoh, and T.

Yatagai, "Non-iterative numerical method for laterally superresolving Fourier

domain optical coherence tomography," Optics Express (2006).

122. Y. Liu, Y. Liang, G. Mu, and X. Zhu, "Deconvolution methods for image

deblurring in optical coherence tomography," J Opt Soc Am A Opt Image Sci Vis

26, 72-77 (2009).

123. X. Qi, M. V. Sivak, G. Isenberg, J. E. Willis, and A. M. Rollins, "Computer-aided

diagnosis of dysplasia in Barrett's esophagus using endoscopic optical coherence

tomography.," Journal of Biomedical Optics 11, 044010 (2006).

124. F. Bazant-Hegemark and N. Stone, "Towards automated classification of clinical

optical coherence tomography data of dense tissues," Lasers in Medical Science

24, 627-638 (2009).

125. D. Koozekanani, K. Boyer, and C. Roberts, "Retinal thickness measurements

from optical coherence tomography using a Markov boundary model," Medical

Imaging, IEEE Transactions on 20, 900-916 (2001).

126. D. C. Fernández, H. M. Salinas, and C. A. Puliafito, "Automated detection of

retinal layer structures on optical coherence tomography images," Optics Express

13, 10200-10216 (2005).

196

127. T. Fabritius, S. Makita, M. Miura, R. Myllyl, and Y. Yasuno, "Automated

segmentation of the macula by optical coherence tomography," Optics Express

17, 15659-15669 (2009).

128. M. K. Garvin, M. D. Abràmoff, X. Wu, S. R. Russell, T. L. Burns, and M. Sonka,

"Automated 3-D Intraretinal Layer Segmentation of Macular Spectral-Domain

Optical Coherence Tomography Images," IEEE Transactions on Medical Imaging

28, 1436-1447 (2009).

129. J. M. Schmitt, A. Knuttel, M. Yadlowsky, and M. A. Eckhaus, "Optical-coherence

tomography of a dense tissue: statistics of attenuation and backscattering," Phys.

Med. Biol. 39, 1705-1720 (1994).

130. D. Levitz, L. Thrane, M. H. Frosz, and P. E. Anderson, "Determination of optical

scattering properties of highly -scattering media in optical coherence tomography

images," Optics Express 12, 249-259 (2004).

131. L. Thrane, M. H. Frosz, T. M. Jorgensen, A. Tycho, H. T. Yura, and P. E.

Andersen, "Extraction of optical scattering parameters and attenuation

compensation in optical coherence tomography images of multilayered tissue

structures," Opt Lett 29, 1641-1643 (2004).

132. F. J. van der Meer, D. J. Faber, D. M. B. Sassoon, M. C. Aalders, G. Pasterkamp,

and T. G. van Leeuwen, "Localized Measurement of Optical Attenuation

Coefficients of Atherosclerotic Plaque Consitutents by Quantitative Optical

Coherence Tomography," IEEE Transactions on Medical Imaging 24, 1369-1375

(2005).

197

133. X. Xu, R. K. Wang, and A. E. Haj, "Investigation of changes in optical

attenuation of bone and neuronal cells in organ culture or three-dimensional

constructs in vitro with optical coherence tomography: relevance to cytochrome

oxidase monitoring," Eur Biophys J 32, 355-362 (2003).

134. D. J. Faber, F. J. van der Meer, and M. C. G. Aalders, "Quantitative measurement

of attenuation coefficients of weakly scattering media using optical coherence

tomography," Optics Express 12, 4353-4365 (2004).

135. C. Xu, J. M. Schmitt, S. G. Carlier, and R. Virmani, "Characterization of

atherosclerosis plaques by measuring both backscattering and attenuation

coefficients in optical coherence tomography," Journal of Biomedical Optics 13,

034003-034008 (2008).

136. U. Morgner, W. Drexler, F. X. Kärtner, X. D. Li, C. Pitris, E. P. Ippen, and J. G.

Fujimoto, "Spectroscopic optical coherence tomography," Opt. Lett. 25, 111-113

(2000).

137. A. Kleber and Y. Rudy, "Basic Mechanisms of Cardiac Impulse Propagation and

Associated Arrhythmias," Physiol. Rev. 84, 431-488 (2004).

138. P. Helm, M. Beg, M. Miller, and R. Winslow, "Measuring and mapping cardiac

fiber and laminar architecture using diffusion tensor MR imaging.," Ann N Y

Acad Sci 1047, 296-307 (2005).

139. W. J. Karlon, J. W. Covell, A. D. McCulloch, J. J. Hunter, and J. H. Omens,

"Automated Measurement of Myofiber Disarray in Transgenic Mice With

Ventricular Expression of ras," The Anatomical Record 252, 612-625 (1998).

198

140. C. M. Ripplinger, V. I. Krinsky, V. P. Nikolski, and I. R. Efimov, "Mechanisms

of unpinning and termination of ventricular tachycardia," Am J Physiol Heart Circ

Physiol 291, H184-H192 (2006).

141. Z. Hu and A. M. Rollins, "Quasi-telecentric optical design of a microscope-

compatible OCT scanner," Optics Express 13, 6407-6415 (2005).

142. N. El Sherif, R. R. Hope, B. J. Scherlag, and R. Lazzara, "Re-entrant ventricular

arrhythmias in the late myocardial infarction period. 2. Patterns of initiation and

termination of re-entry," Circulation 55, 702-719 (1977).

143. N. El Sherif, B. J. Scherlag, R. Lazzara, and R. R. Hope, "Re-entrant ventricular

arrhythmias in the late myocardial infarction period. 1. Conduction characteristics

in the infarction zone," Circulation 55, 686-702 (1977).

144. A. L. Wit, M. A. Allessie, F. I. Bonke, W. Lammers, J. Smeets, and J. J. Fenoglio,

Jr., "Electrophysiologic mapping to determine the mechanism of experimental

ventricular tachycardia initiated by premature impulses. experimental approach

and initial results demonstrating reentrant excitation," Am J Cardiol 49, 166-185:

(1982).

145. A. L. Wit, S. M. Dillon, J. Coromilas, A. E. Saltman, and B. Waldecker,

"Anisotropic reentry in the epicardial border zone of myocardial infarcts," Ann N

Y Acad Sci 591, 86-108 (1990).

146. M. T. Watanabe, M; Fallah-Najmabadi, H., "Cardiac Expression of Polysialylated

NCAM in the Chicken Embryo: Correlation With the Ventricular Conduction

System," Developmental Dynamics 194, 128-141 (1992).

199

147. L. Li, V. Nikolski, D. W. Wallick, I. R. Efimov, and Y. Cheng, "Mechanisms of

enhanced shock-induced arrhythmogenesis in the rabbit heart with healed

myocardial infarction," Am J Physiol Heart Circ Physiol 289(2005).

148. N. Klauke, G. L. Smith, and J. Cooper, "Stimulation of Single Isolated Adult

Ventricular Myocytes within a Low Volume Using a Planar Microelectrode

Array," Biophysical Journal 85, 1766-1774 (2003).

149. D. A. Hooks, I. J. LeGrice, J. D. Harvey, and B. H. Smaill, "Intramural Multisite

Recording of Transmembrane Potential in the Heart," Biophysical Journal 81,

2671-2680 (2001).

150. J. L. Byars, W. M. Smith, R. E. Ideker, and V. G. Fast, "Development of an

Optrode for Intramural Multisite Optical Recordings of Vm in the Heart," Journal

of Cardiovascular Electrophysiology 14, 1196-1202 (2003).

151. M. J. Bishop, B. Rodriguez, J. Eason, J. P. Whiteley, N. Trayanova, and D. J.

Gavaghan, "Synthesis of voltage-sensitive optical signals: application to

panoramic optical mapping.," Biophysical Journal 90, 2938-2945 (2006).

152. C. J. Hyatt, S. F. Mironov, F. J. Vetter, C. W. Zemlin, and A. M. Pertsov, "Optical

Action Potential Upstroke Morphology Reveals Near-Surface Transmural

Propagation Direction " Circulation Research 97, 277-284 (2005).

153. M. J. Bishop, D. J. Gavaghan, N. A. Trayanova, and B. Rodriguez, "Photon

scattering effects in optical mapping of propagation and arrhythmogenesis in the

heart.," Journal of Electrocardiology 40, S75-S80 (2007).

154. C. J. Hyatt, C. W. Zemlin, R. M. Smith, A. Matiukas, A. M. Pertsov, and O.

Bernus, "Reconstructing subsurface electrical wave orientation from cardiac epi-

200

fluorescence recordings: Monte Carlo versus diffusion approximation.," Optics

Express 16, 13758-13772 (2008).

155. O. Bernus, M. Wellner, S. F. Mironov, and A. M. Pertsov, "Simulation of voltage-

sensitive optical signals in three-dimensional slabs of cardiac tissue: application to

transillumination and coaxial imaging methods.," Physics in Medicine and

Biology 50, 215-229 (2005).

156. G. W. Beeler and H. Reuter, "Reconstruction of the action potential of ventricular

myocardial fibres.," Journal of Physiology 268, 177-210 (1977).

157. M. E. Brezinski, "Applications of optical coherence tomography to cardiac and

musculoskeletal diseases: bench to bedside?," J Biomed Opt 12, 051705 (2007).

158. N. A. Patel, X. Li, D. L. Stamper, J. G. Fujimoto, and M. E. Brezinski, "Guidance

of aortic ablation using optical coherence tomography," The International Journal

of Cardiovascular Imaging 19, 171-178 (2003).

159. S. A. Boppart, J. Herrmann, C. Pitris, D. L. Stamper, M. E. Brezinski, and J. G.

Fujimoto, "High-Resolution Optical Coherence Tomography-Guided Laser

Ablation of Surgical Tissue," Journal of Surgical Research 82, 275-284 (1999).

160. B. J. Vakoc, G. J. Tearney, and B. E. Bouma, "Real-time microscopic

visualization of tissue response to laser thermal therapy," Journal of Biomedical

Optics Letters 12, 020501 (2007).

161. M. Ford, Y. Zhou, H. Wang, C. X. Deng, and A. M. Rollins, "Optical coherence

tomography monitoring of cardiac ablation by high-intensity focused ultrasound,"

Proceedings of the SPIE 5686, 432 (2005).

201

162. J. Swartling, S. Palsson, P. Platonov, S. B. Olsson, and S. Andersson-Engels,

"Changes in tissue optical properties due to radio-frequency ablation of

myocardium," Med Biol Eng Comput 41, 403-409 (2003).

163. S. Sato, T. Shimada, M. Ishihara, T. Arai, T. Matsui, A. Kurita, M. Obara, M.

Kikuchi, H. Wakisaka, and H. Ashida, "Laser Ablation Characteristics of

Myocardium Tissue in the UV Spectral Region: An In-vitro Study with Porcine

Myocardium Tissue," in OSA BOSD, (1999).

164. S. Bosman, "Heat-induced structural alterations in myocardium in relation to

changing optical properties," Applied Optics 32, 461-463 (1993).

165. B. Liu, M. Harman, S. Giattina, D. Stamper, Charles Demakis, M. Chilek, S.

Raby, and M. Brezinski, "Characterizing of tissue microstructure with single-

detector polarization-sensitive optical coherence tomography," Applied Optics 45,

4464-4479 (2006).

166. S. D. Giattina, B. K. Courtney, P. R. Herz, M. Harman, S. Shortkroff, D. L.

Stamper, B. Liu, J. G. Fujimoto, and M. E. Brezinski, "Assessment of coronary

plaque collagen with polarization sensitive optical coherence tomography (PS-

OCT)," International Journal of Cardiology 107, 400-409 (2006).

167. Y. Yang, L. Wu, Y. Feng, and R. K. Wang, "Observations of birefringence in

tissues from optic-fibre-based optical coherence tomography," Measurement

Science and Technology 14, 41–46 (2003).

168. B. Holmbom, U. Niislund, A. Eriksson, I. Virtancn, and L.-E. Thornell,

"Comparison of triphenyltetrazolium chloride (TTC) staining versus detection of

202

fibronectin in experimental myocardial infarction," Histochemistry 99, 265-275

(1993).

169. C. Xu, J. M. Schmitt, S. G. Carlier, and R. Virmani, "Characterization of

atherosclerosis plaques by measuring both backscattering and attenuation

coefficients in optical coherence tomography," Journal of Biomedical Optics

13(2008).

170. G. J. Tearney, M. E. Brezinski, J. F. Southern, B. E. Bouma, M. R. Hee, and J. G.

Fujimoto, "Determination of the refractive index of highly scattering human tissue

by optical coherence tomography," Optics Letters 20, 2258-2260 (1995).

171. A. M. Rollins, M. D. Kulkarni, S. Yazdanfar, R. Un-arunyawee, and J. A. Izatt,

"In Vivo Video Rate Optical Coherence Tomography," Optics Express 3, 219-229

(1998).

172. G. Bonanomi, D. Schwartzman, D. Francischelli, K. Hebsgaard, and M. A.

Zenati, "A new device for beating heart bipolar radiofrequency atrial ablation," J

Thorac Cardiovasc Surg 126, 1859-1866 (2003).

173. L. Bretzner and T. Lindeberg, "Feature Tracking with Automatic Selection of

Spatial Scales," Computer Vision and Image Understanding 71, 385-392 (1998).

174. P. Whittaker, S.-m. Zheng, M. J. Patterson, R. A. Kloner, K. E. Daly, and R. A.

Hartman, "Histologic Signatures of Thermal Injury: Applications in

Transmyocardial Laser Revascularization and Radiofrequency Ablation," Lasers

in Surgery and Medicine 27, 305–318 (2000).

203

175. J. M. Cooper, J. L. Sapp, U. Tedrow, C. P. Pellegrin, D. Robinson, L. M. Epstein,

and W. G. Stevenson, "Ablation with an internally irrigated radiofrequency

catheter: Learning how to avoid steam pops," Heart Rhythm 1, 329–333 (2004).

176. C. Grimard, J. Lacotte, F. Hidden-Lucet, G. Duthoit, Y. Gallais, and R. Frank,

"Percutaneous Epicardial Radiofrequency Ablation of Ventricular Arrhythmias

After Failure of Endocardial Approach: A 9-Year Experience," Journal of

Cardiovascular Electrophysiology Ahead of print(2009).

177. E. Sosa and M. Scanavacca, "Epicardial mapping and ablation techniques to

control ventricular tachycardia.," Journal of Cardiovascular Electrophysiology 16,

449-452 (2005).

178. P. Wang, H. Hsia, A. Al-Ahmad, and P. Zei, eds., Ventricular Arrhythmias and

Sudden Cardiac Death: Mechanism, Ablation, and Defibrillation (Blackwell,

Malden, MA, 2008).

179. A. d’Avila, C. Houghtaling, P. Gutierrez, O. Vragovic, J. N. Ruskin, M. E.

Josephson, and V. Y. Reddy, "Catheter Ablation of Ventricular Epicardial Tissue:

A Comparison of Standard and Cooled-Tip Radiofrequency Energy," Circulation

109, 2363-2369 (2004).

180. A. e. d’Avila, R. Splinter, R. H. Svenson, M. Scanavacca, E. Pruitt, J. Kasell, and

E. Sosa, "New Perspectives on Catheter-Based Ablation of Ventricular

Tachycardia Complicating Chagas’ Disease: Experimental Evidence of the

Efficacy of Near Infrared Lasers for Catheter Ablation of Chagas’ VT," Journal of

Interventional Cardiac Electrophysiology 7, 23-38 (2002).

204

181. S. Dixit, N. Narula, D. J. Callans, and F. Marchlinski, "Electroanatomic Mapping

of Human Heart: Epicardial Fat Can Mimic Scar," Journal of Cardiovascular

Electrophysiology 14, 1128 (2003).

182. A. D'Avila, P. Gutierrez, M. Scanavacca, V. Reddy, D. L. Lustgarten, E. Sosa,

and J. A. Ramires, "Effects of radiofrequency pulses delivered in the vicinity of

the coronary arteries: implications for nonsurgical transthoracic epicardial

catheter ablation to treat ventricular tachycardia," Pacing Clin Electrophysiol 25,

1488-1495 (2002).

183. A. d’Avila, "Epicardial catheter ablation of ventricular tachycardia," Heart

Rhythm 5, S73-S75 (2008).

184. H. Ashikaga, T. Sasano, J. Dong, M. M. Zviman, R. Evers, B. Hopenfeld, V.

Castro, R. H. Helm, T. Dickfeld, S. Nazarian, J. K. Donahue, R. D. Berger, H.

Calkins, M. R. Abraham, E. Marbán, A. C. Lardo, E. R. McVeigh, and H. R.

Halperin, "Magnetic resonance-based anatomical analysis of scar-related

ventricular tachycardia: implications for catheter ablation," Circulation Research

101, 939-947 (2007).

185. B. Desjardins, T. Crawford, E. Good, H. Oral, A. Chugh, F. Pelosi, F. Morady,

and F. Bogun, "Infarct architecture and characteristics on delayed enhanced

magnetic resonance imaging and electroanatomic mapping in patients with

postinfarction ventricular arrhythmia," Heart Rhythm 6, 644-651 (2009).

186. C. P. Fleming, K. J. Quan, H. Wang, G. Amit, and A. M. Rollins, "In vitro

characterization of cardiac radiofrequency ablation lesions using optical

coherence tomography," Optics Express 18, 3079-3092 (2010).

205

187. A. F. Low, G. J. Tearney, B. E. Bouma, and I. K. Jang, "Technology Insight:

optical coherence tomography—current status and future development," Nature

clinical practice. Cardiovascular medicine 3, 154-162 (2006).

188. T. Fabritius, S. Makita, Y. Hong, R. Myllyla, and Y. Yasuno, "Automated retinal

shadow compensation of optical coherence tomography images," Journal of

Biomedical Optics Letters 14, 010503 (2009).

189. B. D. Goldberg, N. V. Iftimia, J. E. Bressner, M. B. Pitman, E. Halpern, B. E.

Bouma, and G. J. Tearney, "Automated algorithm for differentiation of human

breast tissue using low coherence interferometry for fine needle aspiration biopsy

guidance," Journal of Biomedical Optics 13, 014014 (2008).

190. K. W. Gossage, T. S. Tkaczyk, J. J. Rodriguez, and J. K. Barton, "Texture

analysis of optical coherence tomography images: feasibility for tissue

classification.," Journal of Biomedical Optics 8, 570-575 (2003).

191. C. P. Fleming, K. J. Quan, and A. M. Rollins, "Towards Guidance of Epicardial

Cardiac Radiofrequency Ablation Therapy using Optical Coherence

Tomography," Journal of Biomedical Optics Accepted(2010).

192. C. P. Fleming, H. Wang, K. J. Quan, and A. M. Rollins, "Real time monitoring of

cardiac radiofrequency ablation lesion formation using a forward imaging optical

coherence tomography catheter," In preparation.

193. A. Ahmad, S. G. Adie, E. J. Chaney, U. Sharma, and S. A. Boppart, "Cross-

correlation-based image acquisition technique for manually-scanned optical

coherence tomography," Optics Express 17, 8125-8136 (2009).

206

194. W. F. Cheong, S. A. Prahl, and A. J. Welch., "A review of the optical properties

of biological tissues," IEEE J. Quantum Electron 26, 2166-2185 (1990).

195. R. Splinter, R. H. Svenson, L. Littmann, C. H. Chuang, J. R. Tuntelder, M.

Thompson, G. P. Tatsis, and M. Keijzer, "Computer simulated light distributions

in myocardial tissue at the Nd-YAG wavelength of 1064," Lasers in Medical

Science 8, 15-21 (1993).

196. P. Whittaker, S. Zheng, M. J. Patterson, R. A. Kloner, K. E. Daly, and R. A.

Hartman, "Histologic signatures of thermal injury: applications in transmyocardial

laser revascularization and radiofrequency ablation," Lasers Surg Med 27, 305-

318 (2000).

197. Y. K. Tao, K. M. Kennedy, and J. A. Izatt, "Velocity-resolved 3D retinal

microvessel imaging using single-pass flow imaging spectral domain optical

coherence tomography.," Optics Express 17, 4177-4188 (2009).

198. A. Mariampillai, B. A. Standish, E. H. Moriyama, M. Khurana, N. R. Munce, M.

K. Leung, J. Jiang, A. Cable, B. C. Wilson, I. A. Vitkin, and V. X. Yang,

"Speckle variance detection of microvasculature using swept-source optical

coherence tomography.," Optics Letters 33, 1530-1532 (2008).

199. A. Codreanu, F. Odille, E. Aliot, P.-Y. Marie, I. Magnin-Poull, M. Andronache,

D. Mandry, W. Djaballah, D. Régent, J. Felblinger, and C. d. Chillou,

"Electroanatomic Characterization of Post-Infarct Scars: Comparison With 3-

Dimensional Myocardial Scar Reconstruction Based on Magnetic Resonance

Imaging," Journal of American College of Cardiology 52, 839-842 (2008).

207

200. H.-N. Pak, C. Hwang, H. E. Lim, J. S. Kim, and Y.-H. Kim, "Hybrid Epicardial

and Endocardial Ablation of Persistent or Permanent Atrial Fibrillation:A New

Approach for Difficult Cases," Journal of Cardiovascular Electrophysiology 18,

917-923 (2007).

201. F. Saremi and M. Tafti, "The Role of Computed Tomography and Magnetic

Resonance Imaging in Ablation Procedures for Treatment of Atrial Fibrillation,"

Seminars in Ultrasound, CT, and MRI 30, 125-126 (2009).

202. I. K. Jang, B. E. Bouma, D. H. Kang, S. J. Park, S. W. Park, K. B. Seung, K. B.

Choi, M. Shishkov, K. Schlendorf, E. Pomerantsev, S. L. Houser, H. T. Aretz, and

G. J. Tearney, "Visualization of coronary atherosclerotic plaques in patients using

optical coherence tomography: comparison with intravascular ultrasound," J Am

Coll Cardiol 39, 604-609 (2002).

203. F. Sullivan, "US Cardiac Rhythm Management Markets. Chapter 6: Cardiac

Ablation Market," (2006).

204. M. W. Jenkins, R. S. Wade, Y. Cheng, I. R. Efimov, and A. M. Rollins, "Optical

coherence tomography imaging of the purkinje network.," Journal of

Cardiovascular Electrophysiology 16, 559-560 (2005).

205. J. E. Roth, J. A. Kozak, S. Yazdanfar, A. M. Rollins, and J. A. Izatt, "Simplified

method for polarization-sensitive optical coherence tomography," Optics Letters

26, 1069-1071 (2001).

206. S. H. Yun, C. Boudoux, G. J. Tearney, and B. E. Bouma, "High-speed

wavelength-swept semiconductor laser with a polygon-scanner-based wavelength

filter," Optics Letters 28, 1981-1983 (2003).

208

207. M. A. Choma, K. Hsu, and J. A. Izatt, "Swept source optical coherence

tomography using an all-fiber 1300-nm ring laser source.," Journal of Biomedical

Optics 10, 44009 (2005).

208. R. Huber, M. Wojtkowski, and J. Fujimoto, "Fourier Domain Mode Locking

(FDML): A new laser operating regime and applications for optical coherence

tomography.," Optics Express 14, 3225-3237 (2006).

209. W. Y. Oh, S. H. Yun, G. J. Tearney, and B. E. Bouma, "115 kHz tuning repetition

rate ultrahigh-speed wavelength-swept semiconductor laser," Optics Express 30,

3159-3161 (2005).

210. D. C. Adler, R. Huber, and J. G. Fujimoto, "Phase-sensitive optical coherence

tomography at up to 370,000 lines per second using buffered Fourier domain

mode-locked lasers," Optics Letters 32, 626-628 (2007).

211. I. Hartl, X. D. Li, C. Chudoba, R. K. Ghanta, T. H. Ko, J. G. Fujimoto, J. K.

Ranka, and R. S. Windeler, "Ultrahigh-resolution optical coherence tomography

using continuum generation in an air silica microstructure optical fiber," Optics

Letters 26, 608-610 (2001).

212. H. Wang, M. W. Jenkins, and A. M. Rollins, "A combined multiple-SLED

broadband light source at 1300 nm for high resolution optical coherence

tomography," Optics Communications 281, 1896-1900 (2008).

213. R. A. Leitgeb, M. Villiger, A. H. Bachmann, L. Steinmann, and T. Lasser,

"Extended focus depth for Fourier domain optical coherence microscopy," Optics

Letters 31, 2450-2452 (2006).

209

214. Y. He and R. K. Wang, "Dynamic optical clearing effect of tissue impregnated

with hyperosmotic agents and studied with optical coherence tomography,"

Journal of Biomedical Optics 9, 200-206 (2004).

215. X. Xu and R. K. Wang, "Synergistic effect of hyperosmotic agents of dimethyl

sulfoxide and glycerol on optical clearing of gastric tissue studied with near

infrared spectroscopy," Physics in Medicine and Biology 49, 457-468 (2004).

210