Disclosures
I have no financial disclosures to make. Imaging & Glaucoma: The New and The Next
Leonard Seibold, M.D. Instructor/Fellow of Ophthalmology Glaucoma & Cataract Service University of Colorado Denver, CO
Ideal Imaging Modality Color Fundus Photography
z Historic gold standard z 1. Differentiate between normal and z But is it the current gold standard??? glaucomatous eyes z Advantages z 2. Detect glaucomatous changes before z Reliable method for documenting disc cupping – (stereophotographs) functional vision loss (Pre-perimetric) z Ability to accurately measure cup/rim distances z No new equipment to buy - Most office settings already have camera z 3. Reliably detect progression of disease z Red-free photos can aid in RNFL loss detection z Useful for detection of change over long follow-up
z Disadvantages z Other beneficial features z Subjective and variable interpretation z Difficulty delineating nerve rim on 2-D photos z Fast, Easy, Applicable to all patients z Cannot quantify RNFL thickness z No reliable normative database to compare to z Not as helpful early on in disease detection
Confocal Scanning Laser Tomography - HRT HRT z 670 nm diode laser beam scans fundus z Intensity of light reflection is a function of z Reflected light from each scan point is measured depth z A pinhole or confocal aperture in front of the detector eliminates scattered light z Retina is dark – prominent z Raster scans of the fundus are built into a 2-D image z Cup is bright - depressed z A series of images are obtained at successive planes of depth z A layered 3-D image is then constructed HRT HRT z 384 x 384 pixel image generated z Several topographic parameters are calculated z 15 x 15 degrees z Compared to normative z Outer disk margin is manually placed database for detection z Reference plane placed at 50 μm below z Followed over time for retinal surface progression z Mean height contour and z Structures above – cup rim area were consistent z Structures below – rim predictors of glaucoma in OHTS
Confocal Scanning Laser HRT Polarimetry - GDx z Advantages z CSLO + Polarimetry z Rapid, simple z No dilation z Based on birefringence of RNFL z 3D ONH images and contour parameters from parallel oriented z Proven detection of disease microtubules z Compared to established normative database z 780 nm polarized laser scans z Disadvantages z Manual tracing of ONH margins concentric circles around the disk z Arbitrary reference plane (changes with IOP, edema, z Retardation of polarized light is cardiac cycle) z Difficulty with tilted/myopic/hyperopic discs proportional to RNFL thickness z Poor RNFL analysis z Cornea and lens contribution must be cancelled out – VCC or ECC
GDx z Advantages z Fast, simple z No dilation z No reference plane – not affected by IOP z Better measure of RNFL z No manual disc tracing z Large normative database z Disadvantages z VCC requires normal macula z Only useful in peripapillary region z “Odd Scans” without appropriate compensation Optical Coherence Tomography (OCT) Schematic z Introduced 1996 z 850nm diode laser z FDA approved cross-sectional ocular imaging z Beam split to reference mirror and eye modality z Interference pattern correspond to thickness and z Used for: depth of reflecting tissues z Evaluation of retinal/macular pathology z Monitoring RNFL thickness in Glaucoma z Corneal/Anterior segment imaging z Time Domain z Three generations z Spectral Domain (Fourier Domain) z Approved 2006
OCT Technology Comparison
Time Domain Spectral Domain z Cross-sectional images allow visualization of Speed 400 A‐scans/sec 29‐55,000 A‐scans/sec each retinal layer Resolution 10‐15 µm 2‐7 µm z Computer segmentation software delineates Image tracking No Yes layer boundaries to calculate thickness Motion Artifact Yes Minimized Ability to detect disease Proven Potentially better z Glaucoma: RNFL and GCC thickness 3D Data sets/modeling No Yes Variability Very good Excellent/?
Time Domain Spectral Domain z Stratus – Zeiss Meditec, Dublin, CA z Cirrus (Zeiss Meditec, Dublin, CA) z Spectralis (Heidelberg Engineering, Heidelberg, Germany) z RTVue (Optovue, Fremont , CA) Study Purpose Methods z 1. Compare agreement of RNFL thickness z 40 normal subjects measurements from Time-Domain OCT z Baseline exam: Va, IOP, ONH exam, HVF 24- (Stratus) and Spectral-Domain OCT (Cirrus, 2, Gonioscopy Spectralis, RTVue) z OCT scans: Stratus, Cirrus, Spectralis, RTVue z 2. Analyze and compare intervisit z All performed same day reproducibility of each machine z All done with same technician z All scans repeated 2-8 weeks later by same technician.
Agreement with Stratus Agreement with Stratus
Mean ± SD STRATU Average 110.10 ± 12.81 Temporal 75.79 ± 13.03 S Superior 133.46 ± 16.71 Nasal 87.57 ± 16.85 Inferior 143.59 ± 19.89 Mean Mean ± SD r difference p‐value CIRRUS Average 98.68 ± 10.89 ‐11.33 0.000 0.911 Temporal 64.88 ± 10.37 ‐10.87 0.000 0.746 Superior 123.52 ± 16.18 ‐10.01 0.000 0.701 Nasal 74.88 ± 10.31 ‐12.48 0.000 0.755 Inferior 132.01 ± 18.91 ‐11.35 0.000 0.861 Mean ± SD SPECTRAL Average 106.59 ± 12.82 ‐3.36 0.001 0.869 IS Temporal 78.54 ± 14.22 3.01 0.034 0.748 Superior 131.4 ± 18.45 ‐2.10 0.222 0.780 Nasal 78.12 ± 13.13 ‐9.35 0.000 0.659 Inferior 137.37 ± 18.95 ‐5.92 0.000 0.854 Mean ± SD RTVUE Average 112.78 ± 13.2 2.81 0.001 0.905 Temporal 88.19 ± 19.35 12.30 0.000 0.490 Superior 135.81 ± 17.51 2.32 0.202 0.752 Nasal 86.81 ± 13.11 ‐0.52 0.768 0.650 Inferior 148.13 ± 20.32 4.82 0.003 0.853 r = Pearson Correlation Cffit
Agreement with Stratus Bland-Altman Analysis Reproducibility Reproducibility
Mean Mean Reproducibility 95% CI CoV 95% CI ICC 95% CI Intervisit Intervisit 8.83 7.21 to 10.45 2.86 2.17 to 3.42 0.94 0.91 to 0.97 Difference 95% CI p‐value Difference 95% CI p‐value STRATUS 15.96 11.44 to 20.47 7.86 4.35 to 10.23 0.85 0.76 to 0.95 18.05 15.44 to 20.65 4.89 3.84 to 5.75 0.87 0.82 to 0.92 Stratus Average 0.37 ‐0.79 to 1.53 0.536 Spectralis Average ‐0.01 ‐1.36 to 1.33 0.983 23.73 19.31 to 28.15 9.28 7.38 to 10.85 0.78 0.72 to 0.84 Temporal ‐0.27 ‐2.24 to 1.71 0.791 Temporal ‐0.87 ‐3.53 to 1.80 0.528 18.31 12.67 to 23.95 4.52 2.62 to 5.82 0.89 0.82 to 0.95
Superior ‐0.01 ‐2.34 to2.31 0.990 Superior ‐0.43 ‐2.07 to 1.20 0.607 8.89 5.75 to 12.02 3.03 1.47 to 4.03 0.92 0.87 to 0.97 CIRRUS 11.91 7.33 to 16.48 5.79 2.82 to 7.69 0.87 0.77 to 0.96 Nasal 1.09 ‐1.63 to 3.81 0.437 Nasal 0.45 ‐0.59 to 1.48 0.401 24.77 11.55 to 37.99 7.10 0.00* ‐ 10.74 0.74 0.55 to 0.93 Inferior 0.70 ‐1.73 to 3.14 0.576 Inferior 0.36 ‐1.37 to 2.09 0.687 17.29 12.84 to 21.73 7.78 5.24 to 9.68 0.70 0.58 to 0.83 18.63 14.37 to 22.88 5.02 3.57 to 6.13 0.88 0.82 to 0.94
Mean Mean 11.72 9.31 to 14.13 3.91 2.81 to 4.77 0.90 0.85 to 0.95 Intervisit Intervisit SPECTRALIS 21.28 16.50 to 26.05 8.79 6.42 to 10.46 0.74 0.65 to 0.82 Difference 95% CI p‐value Difference 95% CI p‐value 14.14 10.44 to 17.84 3.97 1.91 to 5.28 0.93 0.89 to 0.97 8.39 6.86 to 9.91 3.90 2.98 to 4.64 0.95 0.93 to 0.97 Cirrus Average 0.19 ‐1.14 to 1.53 0.777 RTVue Average ‐0.56 ‐1.42 to 0.30 0.209 14.68 9.80 to 19.57 3.68 1.85 ‐ 4.86 0.93 0.88 to 0.98 Temporal 0.33 ‐1.18 to 1.83 0.672 Temporal 2.84 0.20 to 5.47 0.041 RTVUE 6.59 5.59 to 7.59 2.09 1.71 to 2.41 0.97 0.96 to 0.98 Superior 0.75 ‐3.15 to 4.64 0.710 Superior ‐1.24 ‐2.83 to 0.35 0.134 18.96 13.70 to 24.22 7.56 4.79 to 9.57 0.89 0.82 to 0.96 13.09 11.14 to 15.05 3.51 2.79 to 4.11 0.93 0.91 to 0.96 Nasal ‐1.61 ‐3.92 to 0.69 0.178 Nasal ‐0.16 ‐1.98 to 1.66 0.861 15.31 12.24 to 18.38 6.04 4.51 to 7.25 0.86 0.80 to 0.93 Inferior 0.33 ‐2.29 to 2.94 0.807 Inferior ‐0.60 ‐2.03 to 0.83 0.418 12.02 10.00 to 14.03 2.99 2.31 to 3.54 0.96 0.94 to 0.98 All measurements in Micometers (µm), Reproducibility = 2.77 x intervisit within-eye standard deviation (Sw), CoV = Coefficient of Variation, ICC = Intraclass Correlation Coefficient
Scan Quality Spectralis Errors
z Reasons for discarding z Poor image quality z Failure to delineate accurate RNFL/ILM boundaries
z Total scans discarded: 11 z Spectralis: 10 (10/79 = 12.6%) z Cirrus: 1 z RTVue: 0
Conclusions Follow Up Study
z SD-OCT RNFL thicknesses are highly z Can Software upgrades alter RNFL correlated to Stratus measurements? z However, each SD-OCT’s measurements are z Can they improve quality? significantly different than Stratus z Can they reduce segmentation errors? z Physicians should be aware of relationships when following patients z All instruments were highly reproducible between visits z RTVue > Stratus = Cirrus > Spectralis Factors Affecting OCT Measurements Software Improvements
Updates z Operator error Minor software revisions z Lens/Cornea pathology Fix minor bugs, maintain the product, insert basic new functions Provided at no additional cost z Image artifacts (PVD, vessels, etc.) Change in version number after decimal point ( e.g. 1.0 to 1.1) z Decentration Upgrades Major software revisions z Large refractive errors New analysis methods, applications, significant new function upgrades, or expansion of normative data z Segmentation errors May require additional costs Change in version number before decimal point (e.g. 1.0 to 2.0)
Agreement Agreement
Version 4Version 5
Mean ± SD SE Mean ± SD SE Mean r difference p‐value ± ± Average 106.74 13.31 1.57 106.65 13.14 1.55 ‐0.08 0.409 0.998 ± ± Temporal 78.86 16.47 1.94 78.60 16.37 1.93 ‐0.26 0.191 0.995 ± ± Superior 131.51 18.90 2.23 132.49 17.94 2.11 0.97 0.146 0.955 ± ± Nasal 78.13 14.29 1.68 77.93 14.17 1.67 ‐0.19 0.423 0.990 ± ± Inferior 137.46 19.76 2.33 137.42 19.68 2.32 ‐0.04 0.915 0.986
R= Pearson Correlation Coefficient
Segmentation Errors Large Errors – Version 4 z Small: <25% z Medium: 25% - 75% z Large: >75%
Version 4.0 Version 5.1.3 Small 1 1 Medium 3 3 Complete 2 0 Total 6 (8%) 4 (5%)
z Decrease in error rate by 33% Conclusions Large Errors – Version 5
RNFL thicknesses may change after major software upgrades to SD-OCT The difference was not statistically signficant (<1μm)
Is this clinically signficant though? Recent reports show progression of RNFL loss is approximately (-0.72 μm/yr)
Segmentation algorithms were improved
Clinicians should be aware of RNFL changes after software upgrades
Every upgrade is unique and may give different results
Blood flow is the new "it" thing in glaucoma: Background Glaucoma z Functions of the choroid: z Prelaminar ONH blood supply Diabetes
Stratus OCT Cirrus OCT z Blood supply to outer 2/3 of retina Microvascular z Heat sink for retina Disease z Absorption of excess light
Macular Degeneration
Heidelberg Spectralis inverted Heidelberg Spectralis upright
Choroid Thickness
Subfoveal choroid is thickest
Spaide, Am J Ophthal 2009;147;9; 811-815 Background
z There is mounting evidence that ocular blood flow and ocular perfusion pressure relate to the pathogenesis of glaucoma.
z Since vascular dysregulation likely plays a role in the development of glaucoma, and because the choroid supplies the blood to the the prelaminar optic nerve head, it is logical that the choroid may contribute to or be associated with glaucomatous optic neuropathy.
Background Methods z A decrease in subfoveal choroidal thickness (SFCT) z Retrospective case series including 180 eyes with age has been reported using measurements from from 113 patients who underwent SOCT an inverted image on the Spectralis OCT (SOCT, Heidelberg Engineering, Heidelberg, Germany)8 scanning at the Department of Ophthalmology at the University of Colorado Denver between z There is conflicting evidence in the literature about January and September of 2009. the changes in choroidal thickness in eyes with glaucoma; some studies have demonstrated an 7,9 increase in chroidal thickness and others have z SFCT was measured by a masked observer shown a decrease in chroroidal thickness 10, 11 using the calipers within the software of the OCT z In this study we used SOCT to compare subfoveal machine and positioning them from the outer choroidal thickness (SFCT) between patients with glaucoma and ocular hypertension (OHTN) to normal aspect of Bruch’s membrane to the border of the controls. sclera.
Results Results
z SFCT was 32.4μm thinner in patients with glaucoma when compared to normal eyes after multivariate analysis was used to control for age.
z This result was statistically significant when the standard errors were calculated based on each eye (95% confidence interval [CI] -1.41 to -63.39μm; p=0.042),
z This result was no longer statistically significant when the observations were grouped by patient (CI: -80.54 to 15.74μm; p=0.19). Results Conclusions
z Eyes with OHT had an increase in SFCT of z This study demonstrates thinner SFCT in 74.37μm when compared to normal eyes eyes with glaucoma, thicker SFCT in eyes with OHT when compared to normal eyes and after controlling for age. z This result is statistically significant (CI: 3.24 to 145.5μm; p= 0.042). z Our study supports a previous study that demonstrated that SFCT decreases with age8 z Again, statistical significance disappeared when observations were grouped (CI: -30.57 to 179.31μm; p=0.168). z This is the first SOCT imaging study of glaucomatous eyes in which the choroid was measured under direct, in vivo visualization.
The Challenge of Imaging the Conclusions Outflow System of the Eye
z Limitations of our study include small sample size in some subgroups and retrospective design. z Gonioscopy z Gross anatomy z Highly subjective z Decreased SFCT in glaucomatous eyes may possibly serve as a marker for disease and lend new insight into the pathophysiology of glaucoma z Ultrasound z > 80 MHz probes needed z Imaging of the SFCT can be performed easily and may provide an in office metric of choroidal vascular changes in glaucoma patients. z Optical Coherence Tomography z Requires some degree of optical transparency
Two Photon Imaging In two-photon excitation microscopy a Ti-Sapphire laser is normally used and has a pulse width of ~100 femtoseconds and a repetition rate of ~80 MHz.
This allows high photon density and flux required for two photons to be absorbed and is tunable across a wide range of wavelengths.
Added benefits of multiphoton imaging: -Longer wavelengths are scattered to a lesser degree -Lower-energy photons are less likely to cause tissue damage
There are several caveats to using two-photon microscopy: -The technology is very expensive -The hardware is easily damaged Intact Mouse Eye Aqueous Veins In Vivo Imaging THANK YOU!