Measurement of Retinal Vascular Caliber from Optical Coherence Tomography Phase Images
Total Page:16
File Type:pdf, Size:1020Kb
Special Issue Measurement of Retinal Vascular Caliber From Optical Coherence Tomography Phase Images Klemens Fondi,1 Gerold C. Aschinger,2,3 Ahmed M. Bata,1 Piotr A. Wozniak,1 Liang Liao,1 Gerald Seidel,4 Veronika Doblhoff-Dier,2,3 Doreen Schmidl,1,2 Gerhard Garh¨ofer,1 Rene´ M. Werkmeister,1 and Leopold Schmetterer1,2 1Department of Clinical Pharmacology, Medical University of Vienna, Vienna, Austria 2Center for Medical Physics and Biomedical Engineering, Medical University of Vienna, Vienna, Austria 3Institute of Applied Physics, Vienna University of Technology, Vienna, Austria 4Department of Ophthalmology, Medical University of Graz, Graz, Austria Correspondence: Leopold Schmet- PURPOSE. To compare retinal vessel calibers extracted from phase-sensitive optical coherence terer, Department of Clinical Phar- tomography (OCT) images with vessel calibers as obtained from the Retinal Vessel Analyzer macology, Medical University of (RVA). Vienna, W¨ahringer Gurtel¨ 18-20, 1090 Vienna, Austria; METHODS. Data from previously published studies in 13 healthy subjects breathing room air (n leopold.schmetterer@meduniwien. ¼ 214 vessels) and 7 subjects breathing 100% oxygen (n ¼ 101 vessels) were used. Vessel ac.at. calibers from OCT phase images were measured vertically along the optical axis by three Submitted: October 22, 2015 independent graders. The data from RVA fundus images were corrected for magnification to Accepted: December 17, 2015 obtain absolute values. Citation: Fondi K, Aschinger GC, Bata RESULTS. The average vessel diameter as obtained from OCT images during normoxia was AM, et al. Measurement of retinal lower than from RVA images (83.8 6 28.2 lm versus 86.6 6 28.0 lm, P < 0.001). The same vascular caliber from optical coher- phenomenon was observed during 100% oxygen breathing (OCT: 81.0 6 22.4 lm, RVA: 85.5 ence tomography phase images. In- 6 26.0 lm; P ¼ 0.001). Although the agreement between the two methods was generally vest Ophthalmol Vis Sci. high, the difference in individual vessels could be as high as 40%. These differences were 2016;57:OCT121–OCT129. neither dependent on absolute vessel size nor preferably found in specific subjects. DOI:10.1167/iovs.15-18476 Interobserver differences between OCT evaluators were much lower than differences between the techniques. CONCLUSIONS. Extracting vessel calibers from OCT phase images may be an attractive approach to overcome some of the problems associated with fundus imaging. The source of differences in vessel caliber between the two methods remains to be investigated. In addition, it remains unclear whether OCT-based vessel caliber measurement is superior to fundus camera–based imaging in risk stratification for systemic or ocular disease. (ClinicalTrials.gov numbers, NCT00914407, NCT02531399.) Keywords: retinal vessel diameter, humans, retinal perfusion etinal vascular calibers are usually measured from fundus A potential approach to overcome these problems is to use R photographs using digital imaging.1 Most approaches optical coherence tomography (OCT). This technique offers the nowadays follow the formula developed by Hubbard and advantage of easier recording, provides three-dimensional coworkers1 to calculate the central retinal arteriolar equivalent information, and is, at least in depth, independent of magnifi- (CRAE) and the central retinal venular equivalent (CRVE). The cation problems. The measurement of vascular caliber data from dimensionless quotient arteriovenous ratio (AVR) is used in OCT images is, however, not straightforward. In OCT, larger most studies because it is independent of the magnification of retinal vessels cause a characteristic shadowing effect that is the image, which depends on both the optics of the fundus caused by the scattering of light at red blood cells (RBCs). camera and the optical properties of the eye.1 Extraction of data can be done either vertically along the axis of Because abnormalities in retinal vascular calibers are the illuminating beam or horizontally in the retinal plane. In this associated with a wide variety of cardiovascular, ocular, kidney, context, it needs to be considered that the resolution of typical and brain diseases, accurate measurement of retinal vascular commercially available OCT systems is in the order of 5 lm calibers is desired.2 However, some limitations of the currently verticallyand15to20lm horizontally.4 In larger vessels, a available technology prevent the translation of such measure- characteristic shadowing effect caused by the scattering of light ment into clinical praxis: the absolute measurement of vessel at RBCs can impair accurate vessel delineation. It follows that caliber is not possible, measurements are usually done from several different approaches were published to extract caliber one fundus image only and recorded at an undefined time point data from retinal vessels based on OCT.5–8 during the cardiac cycle, pupil dilatation is required, and the In the present study, we set out to measure retinal vascular three-dimensional geometry of the vessel is not taken into caliber from the phase values of the complex OCT signal. More account.2,3 than a decade ago, it was shown that extraction of the phase iovs.arvojournals.org j ISSN: 1552-5783 OCT121 This work is licensed under a Creative Commons Attribution 4.0 International License. Downloaded from iovs.arvojournals.org on 10/02/2021 OCT and Retinal Vascular Caliber IOVS j Special Issue j Vol. 57 j No. 9 j OCT122 from Fourier-domain OCT (FD-OCT) images can be used to contrast blood vessels.9,10 Nowadays these techniques are widely used in OCT angiography and also form the basis for quantitative blood flow measurement.11 This approach was used in the present study to measure vessel diameters and then compare them to the measurements of vessel calibers from fundus images in healthy subjects. METHODS Subjects The data in the present report were obtained from one yet unpublished study and two studies that were published previously.12,13 The studies were undertaken to measure total retinal blood flow and total retinal oxygen extraction in healthy subjects. The study protocols were approved by the Ethics Committee of the Medical University of Vienna and followed the guidelines set forth in the Declaration of Helsinki. All subjects passed a screening examination before the study day that included a physical examination, blood sampling to assess hematologic status and chemistry, a 12-lead electrocardiogram, the measurement of visual acuity, slit lamp biomicroscopy, funduscopy, and the measurement of IOP. Exclusion criteria were ametropia ‡3 diopters, anisometropia ‡3 diopters, other ocular abnormalities, and any clinically relevant illness as judged by the investigators, as well as a blood donation or intake of any medication in the 3 weeks before the study. The participants had to abstain from beverages containing alcohol or caffeine in the 12 hours before the study visit. Protocol The measurements were conducted under dilated pupil conditions using a custom-built dual-beam bidirectional Doppler FD-OCT to measure phase shifts and a Retinal Vessel Analyzer (RVA; Imedos Systems UG, Jena, Germany) to quantify vessel calibers.12,13 Subjects were measured while breathing ambient air (n¼ 13) or 100% oxygen (n ¼ 7, gases for human use; Messer, Vienna, Austria), which was delivered via a partially expanded reservoir bag at atmospheric pressure. The phase of inhaling 100% oxygen lasted 30 minutes and measurements began 15 minutes after the start of the inhalation. Extraction of Retinal Calibers From OCT Phase Images The measurements were performed with a dual-beam Doppler FD-OCT system as described in detail previously.12,14,15 Briefly, two orthogonally polarized probe beams under a known angle Da are used to illuminate the fundus. The system records the FIGURE 1. Sample measurement in a healthy subject. Fundus image (A) spectra of the two channels as a function of frequency using and OCT phase image (B) are acquired at the same position. Data from two identical spectrometers. When the beams fall onto blood fundus images are obtained within the retinal plane, data from phase vessels, the back-reflected light in each channel is Doppler- OCT images are obtained from the depth profile, because OCT provides better resolution in depth than transversally. shifted because of the movement of the RBCs. We have shown that with such a setup it is possible to measure absolute blood flow in particular retinal vessels14 as well as total retinal blood phase image from the channel that showed better phase flow.12 The refractive index of blood (1.37) at the used light contrast between moving RBCs and static tissue. The outer source’s wavelength was included in the calculations.14 In the border of the phase-shifted areas was determined manually by current study, we extracted retinal calibers from these three independent observers (KF, AMB, LL). The retinal vessel measurements (Fig. 1). A total of 60 frames were recorded diameter as obtained from the OCT phase images was defined for each vessel during a measurement period of 5 seconds. as the mean of those seven measurements. Hence, our observation time was longer than one cardiac cycle. Frames 0, 10, 20, 30, 40, 50, and 59 of each recording Extraction of Retinal Calibers From Fundus Images were evaluated, resulting in a total of seven diameter readings for each vessel. In each of these images, the areas with phase The OCT setup is integrated into a commercially available RVA shifts caused by the moving RBCs were evaluated, using the (Imedos Systems UG), and the details of this coupling were Downloaded from iovs.arvojournals.org on 10/02/2021 OCT and Retinal Vascular Caliber IOVS j Special Issue j Vol. 57 j No. 9 j OCT123 FIGURE 2. Bland-Altman plots comparing retinal vessel diameters as obtained with OCT and RVA. Data are presented during both breathing room air and 100% oxygen. Retinal arteries (red) and retinal veins (blue) are presented separately. On the x-axis, the mean value of RVA and OCT data are presented; on the y-axis, the difference between values as obtained with RVA and OCT is shown.