Jpn J Ophthalmol (2016) 60:166–171 DOI 10.1007/s10384-016-0435-3
CLINICAL INVESTIGATION
Comparison of retinal vessel measurements using adaptive optics scanning laser ophthalmoscopy and optical coherence tomography Shigeta Arichika1 • Akihito Uji1 • Sotaro Ooto1 • Yuki Muraoka1 • Nagahisa Yoshimura1
Received: 1 July 2015 / Accepted: 24 January 2016 / Published online: 23 February 2016 Japanese Ophthalmological Society 2016
Abstract Purpose We compared adaptive optics scanning laser ophthalmoscopy (AOSLO) and optical coherence tomography (OCT) vessel caliber measurements. Methods AOSLO videos were acquired from 28 volunteers with healthy eyes. Artery measurements were made 0.5–1 disc diameters away from the optic disc margin. Individual segmented retinal arterial caliber was measured in synchronization with cardiac pulsation and averaged to obtain final horizontal retinal arterial caliber (ACH) and horizontal retinal arterial lumen (ALH). All OCT images were obtained with the Spectralis OCT, a spectral-domain OCT system. Vertical retinal arterial caliber (ACV) and vertical retinal arterial lumen (ALV) were measured on the same artery measured with AOSLO. Measurements made with the two imaging systems were compared. Results Average ACH, measured with AOSLO, was 123.4 ± 11.2 and average ALH was 101.8 ± 10.2 lm. Average ACV, measured with OCT, was 125.5 ± 11.4 and average ALV was 99.1 ± 10.6 lm. Both arterial caliber (r = 0.767, p \ 0.0001) and arterial lumen (r = 0.81, p \ 0.0001) measurements were significantly correlated between imaging modalities. Additionally, ACH and ACV were not significantly different (p = 0.16). However, ALH measurements were significantly higher than ALV measurements (p = 0.03). Conclusions Vessel measurements made with AOSLO and OCT were well correlated. Moreover, plasma is visible
& Akihito Uji [email protected] 1
Department of Ophthalmology and Visual Sciences, Kyoto University Graduate School of Medicine, 54 Shogoin Kawahara-cho, Sakyo-ku, Kyoto 606-8507, Japan
123
and distinguishable from the retinal vessel wall in AOSLO images but not in OCT images. Therefore, AOSLO may measure vessel width more precisely than OCT. Keywords Adaptive optics scanning laser ophthalmoscopy Optical coherence tomography Vessel caliber measurement
Introduction Adaptive optics (AO) technology is well known in the field of astronomical observation. This technology has been recently applied to ophthalmological imaging to obtain ultra-high-resolution retinal images that enable extensive evaluation of photoreceptors [1, 2], the nerve fiber layer [3], retinal blood flow [4–6], retinal pericytes [7], and other vessel wall features [8, 9]. Focusing on the vessel wall, Koch et al. [9] observed morphometric changes with blood pressure changes using an AO camera. The retinal vasculature is the only vessel system in the body that can be directly assessed (through the pupil), and many attempts have been made to better understand it. Many investigators relied upon photography-based measurements made with computer-imaging software Interactive Vessel Analysis (IVAN; University of Wisconsin, Madison, WI, USA). These measurements can accurately evaluate vessel caliber but provide almost no information on the vessel wall. Therefore, newer noninvasive technologies have been used to image the retina and its vessels, including scanning laser Doppler flowmetry (SLDF) [10, 11], optical coherence tomography (OCT) [12], and AO techniques [7]. In 2007, Harazny et al. [10] first used SLDF to assess the retinal vasculature. They measured caliber width on reflection images and lumen width on perfusion
Comparison of retinal vessel measurements using adaptive optics scanning laser…
images. In 2012, Chui et al. [8] directly visualized the retinal vessel wall using adaptive optics scanning laser ophthalmoscopy (AOSLO). In 2013, Muraoka et al. [12] reported that OCT images of vessel wall reflectivity can provide information on wall thickness. Koch et al. [9] analyzed morphometric changes in the vessel wall with changes in blood pressure using an AO camera. To the best of our knowledge, few studies have compared vessel measurements made with different imaging modalities. Muraoka et al. [12] report a correlation between retinal vessel diameter measurements made on OCT images and fundus photographs (using IVAN). However, differences in vessel wall assessments made with different imaging modalities were not examined in that study because the vessel wall cannot be assessed with fundus photographs. Because visualization and evaluation of the retinal vessel wall provides information on common lifestyle-related diseases (e.g., hypertension [13] and cerebrovascular disease [10]), identifying a sensitive marker of retinal vasculature endothelial cell dysfunction would be useful [14]. Therefore, a better understanding of how vessel wall properties compare and contrast between different imaging modalities is important. We compared measurements of retinal arterial caliber width, lumen width, and wall thickness with AOSLO and OCT, both of which can noninvasively image the vessel wall.
Participants and methods This study was approved by the Institutional Review Board and Ethics Committee at Kyoto University Graduate School of Medicine. All study conduct adhered to the tenets of the Declaration of Helsinki. After having the study design and the risks and benefits of participation thoroughly explained, all participants provided written informed consent to participate in this study. Participants Twenty-eight healthy volunteers were enrolled in this study. Individuals were excluded from participation if they had a best-corrected visual acuity \20/25, high myopia (spherical refractive error \-6.0 diopters, axial length [26.0 mm), intraocular pressure [21 mmHg, any ocular disease, or if if they were pregnant or had any systemic disease (e.g., systemic hypertension). Retinal vessel imaging All participants underwent AOSLO video imaging with a prototype system (Canon Inc., Tokyo, Japan), specifically
167
designed for our laboratory. One eye from each subject was dilated with 1 drop of 0.5 % tropicamide and 0.5 % phenylephrine hydrochloride before video acquisition. One eye was selected as the study eye. All AOSLO imaging was obtained while participants were seated, and each session took approximately 15 min to complete. AOSLO comprised an AO system, a high-resolution confocal SLO imaging system, and a wide-field imaging subsystem specifically designed for this study [15]. The light-source wavelength was 840 nm and lateral resolution 2.0 lm. All AOSLO videos had a 10-s duration, were recorded at a scan rate of 32 frames/s, and imaged a 0.34-mm 9 0.34mm area. Images were acquired while the optical focus was on the layer in which the vessel wall was optimally visualized. As previously reported [16], actual retinal dimensions were calculated from degrees, using each individual’s axial length, measured with an optical biometer (IOL Master; Carl Zeiss Meditec, Dublin, CA, USA), and AOSLO retinal image analyzer software (ARIA; Canon Inc., Tokyo, Japan). This software was specifically designed for our prototype AOSLO. The AOSLO measurements were coordinated with cardiac pulsation by synchronizing images with pulse oximetry pulse recordings obtained by attaching a pulse oximeter (Oxypal Neo; Nihon Kohden, Tokyo, Japan) to the individual’s earlobe. One cardiac cycle was divided into five segments, and vessel caliber measurements were made during each of these five segments [5]. The mean of the five measurements was defined as the average AOSLO-based value. Imaging with OCT was performed after AOSLO video acquisition. Blood pressure was measured three times to confirm that the participant was not hypertensive. This ensured that the influence of hypertension on vessel caliber would not confound our results. All examinations took place between 3:00 p.m. and 6:00 p.m. (before dinner) to minimize the influence of diurnal variations. Horizontal measurements can be obtained from AOSLO imaging. In contrast, vertical measurements can be obtained from OCT imaging. For simplicity, all AOSLO horizontal retinal arterial caliber width (ACH) measurements were made at the outer surface of the vessel wall. Additionally, all horizontal retinal arterial lumen width (ALH) measurements were made at the inner surface of the vessel wall. Similarly, vertical retinal arterial caliber width (ACV) measurements were made at the outer surface of the vessel wall, and vertical retinal arterial lumen width (ALV) measurements were made at the inner surface of the vessel wall. Wall thickness measured horizontally with AOSLO was defined as WTH and wall-to-lumen ratio as WLRH. Wall thickness measured vertically with OCT was defined as WTV and wall-to-lumen ratio as WLRV. WLRV was determined as wall thickness divided by lumen width.
123
168
Retinal arterial vessel measurements Adaptive optics scanning laser ophthalmoscopy The Automated Retinal Image Analyser (ARIA) software was originally developed to use with AOSLO images to obtain vessel measurements. Raw videos were corrected for scanning distortions and eye motion using the scan line warping method [17, 18]. An averaged image was then created using stabilized frames within each specific portion of the cardiac cycle. Next, an edge-preserving filter was applied to the frame of the averaged image. Control points along retinal arterial axes and walls were manually set to detect retinal arterial wall borders along the running direction of the vessel. Natural spline interpolation was then used to determine vessel wall borders using the control points. Finally, ACH and ALH along the running direction of the retinal artery were measured. At least
S. Arichika et al.
150 lm of vessel length was examined (Fig. 1), with a measurement interval of 6 lm along the length of the vessel, and at least 25 consecutive points were measured. Sixty-four images were averaged to obtain the final image used in analyses. All vessels examined were within zone B, the annular area within 0.5–1 disc diameters away from the optic disc margin [19]. All arterial caliber and lumen width measurements were independently made by two retinal specialists who were masked to all other clinical information. Optical coherence tomography All OCT examinations were performed with the Spectralis HRA ? OCT (Heidelberg Engineering, Heidelberg, Germany) using the methods of Muraoka et al. [12]. The light source wavelength was 870 nm, lateral resolution 6.0 lm, and axial resolution 3.9 lm. A 3.46-mm circular scan focused on the optic disc center was obtained in each eye using a scan rate of 40,000 A-scans/s. The final scan used in analyses was the average of 100 individual scans [12] (Fig. 2). The artery measured using AOSLO was also measured with OCT circular scans obtained 1.73 mm from the optic disc center. Data analyses All values are presented as mean ± standard deviation (SD), where applicable. Intraclass correlation coefficients (ICC) were used to evaluate interdevice and interobserver agreement. Bivariate correlations were analyzed using Pearson correlation coefficients. Paired t tests were used to examine the statistical significance of differences between AOSLO- and OCT-based measurements. All analyses were performed using StatView (Version 5.0; SAS Institute, Cary, NC, USA), except for ICC calculations, which were performed using SPSS statistical software (SPSS, Inc., Chicago, IL, USA). Statistical significance was defined as p \ 0.05.
Results Fig. 1 Adaptive optics scanning laser ophthalmoscopy (AOSLO) measurements of vessel caliber. a Color photograph of a healthy right eye. b AOSLO image of the area in the white square on the photograph in part a. Arterial walls are clearly visible in the AOSLO image. Scale bar 100 lm. c Magnified image of the white square in part b. The arrow represents the axial reflex. Scale bar 100 lm. d The magnified image in part c with segmentation lines added. Retinal arterial wall borders were detected along the vessel’s running direction. Control points for the retinal arterial axes and walls were manually placed. Natural spline interpolation was then used to determine vessel wall borders. The green and blue lines represent vessel wall borders and arterial axis, respectively
123
Participant characteristics are summarized in Table 1. Average ACH and ALH, measured with AOSLO, were 123.4 ± 11.2 and 101.8 ± 10.2 lm, respectively. The mean coefficient of variation for the five measurements was 0.9 %, 1.1 %, 6.2 %, and 5.5 % for caliber width, lumen width, WLRH, and WTH, respectively. Average ACV and ALV were 125.5 ± 11.4 and 99.1 ± 10.6 lm, respectively, as measured with OCT. Significant correlations were found between ACH and ACV (r = 0.767, p \ 0.0001) and between ALH and ALV (r = 0.81,
Comparison of retinal vessel measurements using adaptive optics scanning laser…
169
Fig. 2 Optical coherence tomography (OCT)-based vessel measurements. a Infrared fundus image showing OCT scan-circle placement (circular green arrow). Scan-circle diameter is 3.46 mm and focused on the optic disc center. Images in this figure and in Fig. 1 were obtained from the same individual. b Retinal cross-sectional OCT image corresponding to the green arrow in part a. Scale bar 200 lm.
c Magnified view of the white square in the OCT image shown in part b. Caliber width was defined as the distance between the two outermost vessel hyperreflectivities (red arrow). Lumen width was defined as the distance between the two innermost vessel hyperreflectivities (yellow arrow). Scale bar 200 lm
Table 1 Participant systemic and ocular characteristics
WTH and WTV (r = 0.29, p = 0.14). Both imaging modalities had very high interobserver agreement, with all ICCs C 0.93 (AOSLO: ACH = 0.986, ALH = 0.938; OCT: ACV = 0.987, ALV = 0.996).
Variable
Statistics
Sex (male/female)
12/16
Age (years)
31.1 ± 9.2
Body mass index
21.4 ± 2.5
Smoking status (never/past/current)
26/1/1
Discussion
Systolic
108.8 ± 11.1 (82–126.7)
Diastolic
64.1 ± 8.1 (50.0–76.7)
Caliber width, lumen width, and WLR were highly correlated between horizontal AOSLO and vertical OCT measurements. However, horizontal (AOSLO) and vertical (OCT) WT measurements were not significantly correlated. Vessel caliber is most commonly measured on fundus photographs in the clinical setting using IVAN. However, photographs only visualize vessel caliber, which represents a moving blood column, whereas the surrounding thin, transparent plasma edge stream cannot be seen. Therefore, photograph-based vessel measurements slightly underestimate true vessel lumen width [20]. Lumen width measurements made on OCT images are also likely to be underestimated, because such measurements do not include the plasma layer against the vessel wall. Vessel OCT measurements rely on outer and inner vessel wall
Blood pressure (mmHg)
Intraocular pressure (mmHg)
13.5 ± 3.1 (8.3–20.0)
Heart rate (beats/min)
71.6 ± 13.5
Axial length (mm)
25.1 ± 0.8
Data presented as mean ± standard deviation, where applicable
p \ 0.0001; Fig. 3). Additionally, ACH and ACV were not significantly different (p = 0.16). However, ALH measurements were significantly higher than ALV measurements (p = 0.03). Mean WTH and WLRH were 21.6 ± 2.87 lm and 0.21 ± 0.03, respectively. Mean WTV and WLRV were 26.5 ± 3.27 lm and 0.27 ± 0.04, respectively. There was a significant correlation between WLRH and WLRV (r = 0.45, p = 0.02) but not between
123
170
S. Arichika et al.
Fig. 3 Correlations between vessel measurements made with different imaging modalities. Vessel measurements (caliber width and lumen width) made with adaptive optics scanning laser
ophthalmoscopy (AOSLO) and optical coherence tomography (OCT) were significantly correlated
hyperreflectivity, where projected OCT light hits perpendicular to the vessel [12]. Cimalla et al. [21] speculate that vessel wall hyperreflectivity results from a combination of shear-induced red blood cell alignment within vessels and vessel geometry. Therefore, when using OCT images to evaluate vessel wall thickness, caution should be taken because the plasma stratum may be unintentionally included in the measurements. Interestingly, we found no correlation between AOSLO and OCT measurements of wall thickness, possibly because the proportion of plasma stratum relative to vessel caliber or lumen is inconsistent. These OCT limitations do not apply to AOSLO because this system is able to directly image the vessel wall. Although caliber width measurements with the two imaging modalities were strongly correlated, AOSLO measurements of lumen width were significantly higher than OCT measurements. This suggests that OCT measurements include the thickness of the plasma stratum. Muraoka et al. [12] successfully made venous wall measurements with OCT. However, the proportion of plasma stratum thickness in these measurements should not be ignored. This is especially true because even high-resolution AOSLO systems with an offset mode [8] do not always reliably image very thin venous walls. Our study had several limitations. First, the exact target areas of AOSLO and OCT measurements were slightly different. All OCT measurements were made 1.73 mm from the center of the optic disc, slightly inside zone B (0.5–1 disc diameters from the optic disc margin), but the IVAN system used AOSLO measurements made 1.80 mm from the optic disc center. It is likely that this slight difference only negligibly influenced our results because the AOSLO imaging area was a 0.34-mm 9 0.34-mm square that included the OCT imaging circle. Second, because the
OCT scan circle crosses various vessels around the disc, OCT measurements were calculated using one point for each vessel. In contrast, AOSLO measurements were the average of multiple measurements taken every 6 lm along the same vessel. Third, OCT images provided vertical vascular information, while AOSLO images provided horizontal vascular information. Lastly, the shape of retinal vessels was assumed to be completely round, but this is not always the case. Muraoka et al. [12] previously reported a significant correlation between lumen width measured with OCT and vessel width measured with fundus photographs. Additionally, we found an ICC of 0.759 for AOSLO and OCT caliber width measurements and an ICC of 0.784 for AOSLO and OCT lumen width measurements. This supports the idea that there is little physiological difference between horizontal and vertical lumen width. As shown in Fig. 3, significant correlations were found between ACH and ACV and between ALH and ALV. However, vessels of one volunteer showed a discrepancy between AOSLO and OCT measurements. This may have occurred if the configuration of the retinal artery was not round. To the best of our knowledge, there are no prior comparisons of horizontal and vertical vessel dimensions in humans, and it may have been that this volunteer had a wider horizontal than vertical caliber and lumen. In conclusion, AOSLO- and OCT-based vessel measurements were well correlated. Moreover, unlike OCT images, AOSLO images show plasma and directly visualize the retinal vessel wall. Thus, AOSLO has the potential to measure vessel width more precisely than OCT.
123
Acknowledgments This work was supported, in part, by the Innovative Techno-Hub for Integrated Medical Bio-Imaging of the Project for Developing Innovation Systems, from the Ministry of Education, Culture, Sports, Science and Technology (MEXT) in Japan.
Comparison of retinal vessel measurements using adaptive optics scanning laser… Conflicts of interest S. Arichika, None; A. Uji, None; S. Ooto, None; Y. Muraoka, None; N. Yoshimura, Financial support (Topcon Corporation, Nidek, Canon), Consultant (Nidek).
References 1. Scoles D, Sulai YN, Langlo CS, Fishman GA, Curcio CA, Carroll J, et al. In vivo imaging of human cone photoreceptor inner segments. Invest Ophthalmol Vis Sci. 2014;55:4244–51. 2. Roorda A, Williams DR. The arrangement of the three cone classes in the living human eye. Nature. 1999;397:520–2. 3. Takayama K, Ooto S, Hangai M, Ueda-Arakawa N, Yoshida S, Akagi T, et al. High-resolution imaging of retinal nerve fiber bundles in glaucoma using adaptive optics scanning laser ophthalmoscopy. Am J Ophthalmol. 2013;155:870–81. 4. Tam J, Tiruveedhula P, Roorda A. Characterization of single-file flow through human retinal parafoveal capillaries using an adaptive optics scanning laser ophthalmoscope. Biomed Opt Express. 2011;2:781–93. 5. Arichika S, Uji A, Murakami T, Unoki N, Yoshitake S, Dodo Y, et al. Retinal hemorheological characterization of early-stage diabetic retinopathy using adaptive optics scanning laser ophthalmoscopy. Invest Ophthalmol Vis Sci. 2014;55:8513–22. 6. Bedggood P, Metha A. Direct visualization and characterization of erythrocyte flow in human retinal capillaries. Biomed Opt Express. 2012;3:3264–77. 7. Schallek J, Geng Y, Nguyen H, Williams DR. Morphology and topography of retinal pericytes in the living mouse retina using in vivo adaptive optics imaging and ex vivo characterization. Invest Ophthalmol Vis Sci. 2013;54:8237–50. 8. Chui TY, Vannasdale DA, Burns SA. The use of forward scatter to improve retinal vascular imaging with an adaptive optics scanning laser ophthalmoscope. Biomed Opt Express. 2012;3:2537–49. 9. Koch E, Rosenbaum D, Brolly A, Sahel JA, Chaumet-Riffaud P, Girerd X, et al. Morphometric analysis of small arteries in the human retina using adaptive optics imaging: relationship with blood pressure and focal vascular changes. J Hypertens. 2014;32:890–8. 10. Harazny JM, Ritt M, Baleanu D, Ott C, Heckmann J, Schlaich MP, et al. Increased wall:lumen ratio of retinal arterioles in male patients with a history of a cerebrovascular event. Hypertension. 2007;50:623–9.
171
11. Rizzoni D, Porteri E, Duse S, De Ciuceis C, Rosei CA, La Boria E, et al. Relationship between media-to-lumen ratio of subcutaneous small arteries and wall-to-lumen ratio of retinal arterioles evaluated noninvasively by scanning laser Doppler flowmetry. J Hypertens. 2012;30:1169–75. 12. Muraoka Y, Tsujikawa A, Kumagai K, Akiba M, Ogino K, Murakami T, et al. Age- and hypertension-dependent changes in retinal vessel diameter and wall thickness: an optical coherence tomography study. Am J Ophthalmol. 2013;156:706–14. 13. Ritt M, Harazny JM, Ott C, Schlaich MP, Schneider MP, Michelson G, et al. Analysis of retinal arteriolar structure in never-treated patients with essential hypertension. J Hypertens. 2008;26:1427–34. 14. Cuspidi C, Sala C. Retinal wall-to-lumen ratio: a new marker of endothelial function? J Hypertens. 2011;29:33–5. 15. Arichika S, Uji A, Ooto S, Miyamoto K, Yoshimura N. Adaptive optics-assisted identification of preferential erythrocyte aggregate pathways in the human retinal microvasculature. PLoS ONE. 2014;9:e89679. 16. Bennett AG, Rudnicka AR, Edgar DF. Improvements on Littmann’s method of determining the size of retinal features by fundus photography. Graefes Arch Clin Exp Ophthalmol. 1994;232:361–7. 17. Arichika S, Uji A, Hangai M, Ooto S, Yoshimura N. Noninvasive and direct monitoring of erythrocyte aggregates in human retinal microvasculature using adaptive optics scanning laser ophthalmoscopy. Invest Ophthalmol Vis Sci. 2013;54:4394–402. 18. Uji A, Hangai M, Ooto S, Takayama K, Arakawa N, Imamura H, et al. The source of moving particles in parafoveal capillaries detected by adaptive optics scanning laser ophthalmoscopy. Invest Ophthalmol Vis Sci. 2012;53:171–8. 19. Hubbard LD, Brothers RJ, King WN, Clegg LX, Klein R, Cooper LS, et al. Methods for evaluation of retinal microvascular abnormalities associated with hypertension/sclerosis in the Atherosclerosis Risk in Communities Study. Ophthalmology. 1999;106:2269–80. 20. Baleanu D, Ritt M, Harazny J, Heckmann J, Schmieder RE, Michelson G. Wall-to-lumen ratio of retinal arterioles and arterioleto-venule ratio of retinal vessels in patients with cerebrovascular damage. Invest Ophthalmol Vis Sci. 2009;50:4351–9. 21. Cimalla P, Walther J, Mittasch M, Koch E. Shear flow-induced optical inhomogeneity of blood assessed in vivo and in vitro by spectral domain optical coherence tomography in the 1.3 lm wavelength range. J Biomed Opt. 2011;16:116020.
123
CLINICAL INVESTIGATION
Comparison of retinal vessel measurements using adaptive optics scanning laser ophthalmoscopy and optical coherence tomography Shigeta Arichika1 • Akihito Uji1 • Sotaro Ooto1 • Yuki Muraoka1 • Nagahisa Yoshimura1
Received: 1 July 2015 / Accepted: 24 January 2016 / Published online: 23 February 2016 Japanese Ophthalmological Society 2016
Abstract Purpose We compared adaptive optics scanning laser ophthalmoscopy (AOSLO) and optical coherence tomography (OCT) vessel caliber measurements. Methods AOSLO videos were acquired from 28 volunteers with healthy eyes. Artery measurements were made 0.5–1 disc diameters away from the optic disc margin. Individual segmented retinal arterial caliber was measured in synchronization with cardiac pulsation and averaged to obtain final horizontal retinal arterial caliber (ACH) and horizontal retinal arterial lumen (ALH). All OCT images were obtained with the Spectralis OCT, a spectral-domain OCT system. Vertical retinal arterial caliber (ACV) and vertical retinal arterial lumen (ALV) were measured on the same artery measured with AOSLO. Measurements made with the two imaging systems were compared. Results Average ACH, measured with AOSLO, was 123.4 ± 11.2 and average ALH was 101.8 ± 10.2 lm. Average ACV, measured with OCT, was 125.5 ± 11.4 and average ALV was 99.1 ± 10.6 lm. Both arterial caliber (r = 0.767, p \ 0.0001) and arterial lumen (r = 0.81, p \ 0.0001) measurements were significantly correlated between imaging modalities. Additionally, ACH and ACV were not significantly different (p = 0.16). However, ALH measurements were significantly higher than ALV measurements (p = 0.03). Conclusions Vessel measurements made with AOSLO and OCT were well correlated. Moreover, plasma is visible
& Akihito Uji [email protected] 1
Department of Ophthalmology and Visual Sciences, Kyoto University Graduate School of Medicine, 54 Shogoin Kawahara-cho, Sakyo-ku, Kyoto 606-8507, Japan
123
and distinguishable from the retinal vessel wall in AOSLO images but not in OCT images. Therefore, AOSLO may measure vessel width more precisely than OCT. Keywords Adaptive optics scanning laser ophthalmoscopy Optical coherence tomography Vessel caliber measurement
Introduction Adaptive optics (AO) technology is well known in the field of astronomical observation. This technology has been recently applied to ophthalmological imaging to obtain ultra-high-resolution retinal images that enable extensive evaluation of photoreceptors [1, 2], the nerve fiber layer [3], retinal blood flow [4–6], retinal pericytes [7], and other vessel wall features [8, 9]. Focusing on the vessel wall, Koch et al. [9] observed morphometric changes with blood pressure changes using an AO camera. The retinal vasculature is the only vessel system in the body that can be directly assessed (through the pupil), and many attempts have been made to better understand it. Many investigators relied upon photography-based measurements made with computer-imaging software Interactive Vessel Analysis (IVAN; University of Wisconsin, Madison, WI, USA). These measurements can accurately evaluate vessel caliber but provide almost no information on the vessel wall. Therefore, newer noninvasive technologies have been used to image the retina and its vessels, including scanning laser Doppler flowmetry (SLDF) [10, 11], optical coherence tomography (OCT) [12], and AO techniques [7]. In 2007, Harazny et al. [10] first used SLDF to assess the retinal vasculature. They measured caliber width on reflection images and lumen width on perfusion
Comparison of retinal vessel measurements using adaptive optics scanning laser…
images. In 2012, Chui et al. [8] directly visualized the retinal vessel wall using adaptive optics scanning laser ophthalmoscopy (AOSLO). In 2013, Muraoka et al. [12] reported that OCT images of vessel wall reflectivity can provide information on wall thickness. Koch et al. [9] analyzed morphometric changes in the vessel wall with changes in blood pressure using an AO camera. To the best of our knowledge, few studies have compared vessel measurements made with different imaging modalities. Muraoka et al. [12] report a correlation between retinal vessel diameter measurements made on OCT images and fundus photographs (using IVAN). However, differences in vessel wall assessments made with different imaging modalities were not examined in that study because the vessel wall cannot be assessed with fundus photographs. Because visualization and evaluation of the retinal vessel wall provides information on common lifestyle-related diseases (e.g., hypertension [13] and cerebrovascular disease [10]), identifying a sensitive marker of retinal vasculature endothelial cell dysfunction would be useful [14]. Therefore, a better understanding of how vessel wall properties compare and contrast between different imaging modalities is important. We compared measurements of retinal arterial caliber width, lumen width, and wall thickness with AOSLO and OCT, both of which can noninvasively image the vessel wall.
Participants and methods This study was approved by the Institutional Review Board and Ethics Committee at Kyoto University Graduate School of Medicine. All study conduct adhered to the tenets of the Declaration of Helsinki. After having the study design and the risks and benefits of participation thoroughly explained, all participants provided written informed consent to participate in this study. Participants Twenty-eight healthy volunteers were enrolled in this study. Individuals were excluded from participation if they had a best-corrected visual acuity \20/25, high myopia (spherical refractive error \-6.0 diopters, axial length [26.0 mm), intraocular pressure [21 mmHg, any ocular disease, or if if they were pregnant or had any systemic disease (e.g., systemic hypertension). Retinal vessel imaging All participants underwent AOSLO video imaging with a prototype system (Canon Inc., Tokyo, Japan), specifically
167
designed for our laboratory. One eye from each subject was dilated with 1 drop of 0.5 % tropicamide and 0.5 % phenylephrine hydrochloride before video acquisition. One eye was selected as the study eye. All AOSLO imaging was obtained while participants were seated, and each session took approximately 15 min to complete. AOSLO comprised an AO system, a high-resolution confocal SLO imaging system, and a wide-field imaging subsystem specifically designed for this study [15]. The light-source wavelength was 840 nm and lateral resolution 2.0 lm. All AOSLO videos had a 10-s duration, were recorded at a scan rate of 32 frames/s, and imaged a 0.34-mm 9 0.34mm area. Images were acquired while the optical focus was on the layer in which the vessel wall was optimally visualized. As previously reported [16], actual retinal dimensions were calculated from degrees, using each individual’s axial length, measured with an optical biometer (IOL Master; Carl Zeiss Meditec, Dublin, CA, USA), and AOSLO retinal image analyzer software (ARIA; Canon Inc., Tokyo, Japan). This software was specifically designed for our prototype AOSLO. The AOSLO measurements were coordinated with cardiac pulsation by synchronizing images with pulse oximetry pulse recordings obtained by attaching a pulse oximeter (Oxypal Neo; Nihon Kohden, Tokyo, Japan) to the individual’s earlobe. One cardiac cycle was divided into five segments, and vessel caliber measurements were made during each of these five segments [5]. The mean of the five measurements was defined as the average AOSLO-based value. Imaging with OCT was performed after AOSLO video acquisition. Blood pressure was measured three times to confirm that the participant was not hypertensive. This ensured that the influence of hypertension on vessel caliber would not confound our results. All examinations took place between 3:00 p.m. and 6:00 p.m. (before dinner) to minimize the influence of diurnal variations. Horizontal measurements can be obtained from AOSLO imaging. In contrast, vertical measurements can be obtained from OCT imaging. For simplicity, all AOSLO horizontal retinal arterial caliber width (ACH) measurements were made at the outer surface of the vessel wall. Additionally, all horizontal retinal arterial lumen width (ALH) measurements were made at the inner surface of the vessel wall. Similarly, vertical retinal arterial caliber width (ACV) measurements were made at the outer surface of the vessel wall, and vertical retinal arterial lumen width (ALV) measurements were made at the inner surface of the vessel wall. Wall thickness measured horizontally with AOSLO was defined as WTH and wall-to-lumen ratio as WLRH. Wall thickness measured vertically with OCT was defined as WTV and wall-to-lumen ratio as WLRV. WLRV was determined as wall thickness divided by lumen width.
123
168
Retinal arterial vessel measurements Adaptive optics scanning laser ophthalmoscopy The Automated Retinal Image Analyser (ARIA) software was originally developed to use with AOSLO images to obtain vessel measurements. Raw videos were corrected for scanning distortions and eye motion using the scan line warping method [17, 18]. An averaged image was then created using stabilized frames within each specific portion of the cardiac cycle. Next, an edge-preserving filter was applied to the frame of the averaged image. Control points along retinal arterial axes and walls were manually set to detect retinal arterial wall borders along the running direction of the vessel. Natural spline interpolation was then used to determine vessel wall borders using the control points. Finally, ACH and ALH along the running direction of the retinal artery were measured. At least
S. Arichika et al.
150 lm of vessel length was examined (Fig. 1), with a measurement interval of 6 lm along the length of the vessel, and at least 25 consecutive points were measured. Sixty-four images were averaged to obtain the final image used in analyses. All vessels examined were within zone B, the annular area within 0.5–1 disc diameters away from the optic disc margin [19]. All arterial caliber and lumen width measurements were independently made by two retinal specialists who were masked to all other clinical information. Optical coherence tomography All OCT examinations were performed with the Spectralis HRA ? OCT (Heidelberg Engineering, Heidelberg, Germany) using the methods of Muraoka et al. [12]. The light source wavelength was 870 nm, lateral resolution 6.0 lm, and axial resolution 3.9 lm. A 3.46-mm circular scan focused on the optic disc center was obtained in each eye using a scan rate of 40,000 A-scans/s. The final scan used in analyses was the average of 100 individual scans [12] (Fig. 2). The artery measured using AOSLO was also measured with OCT circular scans obtained 1.73 mm from the optic disc center. Data analyses All values are presented as mean ± standard deviation (SD), where applicable. Intraclass correlation coefficients (ICC) were used to evaluate interdevice and interobserver agreement. Bivariate correlations were analyzed using Pearson correlation coefficients. Paired t tests were used to examine the statistical significance of differences between AOSLO- and OCT-based measurements. All analyses were performed using StatView (Version 5.0; SAS Institute, Cary, NC, USA), except for ICC calculations, which were performed using SPSS statistical software (SPSS, Inc., Chicago, IL, USA). Statistical significance was defined as p \ 0.05.
Results Fig. 1 Adaptive optics scanning laser ophthalmoscopy (AOSLO) measurements of vessel caliber. a Color photograph of a healthy right eye. b AOSLO image of the area in the white square on the photograph in part a. Arterial walls are clearly visible in the AOSLO image. Scale bar 100 lm. c Magnified image of the white square in part b. The arrow represents the axial reflex. Scale bar 100 lm. d The magnified image in part c with segmentation lines added. Retinal arterial wall borders were detected along the vessel’s running direction. Control points for the retinal arterial axes and walls were manually placed. Natural spline interpolation was then used to determine vessel wall borders. The green and blue lines represent vessel wall borders and arterial axis, respectively
123
Participant characteristics are summarized in Table 1. Average ACH and ALH, measured with AOSLO, were 123.4 ± 11.2 and 101.8 ± 10.2 lm, respectively. The mean coefficient of variation for the five measurements was 0.9 %, 1.1 %, 6.2 %, and 5.5 % for caliber width, lumen width, WLRH, and WTH, respectively. Average ACV and ALV were 125.5 ± 11.4 and 99.1 ± 10.6 lm, respectively, as measured with OCT. Significant correlations were found between ACH and ACV (r = 0.767, p \ 0.0001) and between ALH and ALV (r = 0.81,
Comparison of retinal vessel measurements using adaptive optics scanning laser…
169
Fig. 2 Optical coherence tomography (OCT)-based vessel measurements. a Infrared fundus image showing OCT scan-circle placement (circular green arrow). Scan-circle diameter is 3.46 mm and focused on the optic disc center. Images in this figure and in Fig. 1 were obtained from the same individual. b Retinal cross-sectional OCT image corresponding to the green arrow in part a. Scale bar 200 lm.
c Magnified view of the white square in the OCT image shown in part b. Caliber width was defined as the distance between the two outermost vessel hyperreflectivities (red arrow). Lumen width was defined as the distance between the two innermost vessel hyperreflectivities (yellow arrow). Scale bar 200 lm
Table 1 Participant systemic and ocular characteristics
WTH and WTV (r = 0.29, p = 0.14). Both imaging modalities had very high interobserver agreement, with all ICCs C 0.93 (AOSLO: ACH = 0.986, ALH = 0.938; OCT: ACV = 0.987, ALV = 0.996).
Variable
Statistics
Sex (male/female)
12/16
Age (years)
31.1 ± 9.2
Body mass index
21.4 ± 2.5
Smoking status (never/past/current)
26/1/1
Discussion
Systolic
108.8 ± 11.1 (82–126.7)
Diastolic
64.1 ± 8.1 (50.0–76.7)
Caliber width, lumen width, and WLR were highly correlated between horizontal AOSLO and vertical OCT measurements. However, horizontal (AOSLO) and vertical (OCT) WT measurements were not significantly correlated. Vessel caliber is most commonly measured on fundus photographs in the clinical setting using IVAN. However, photographs only visualize vessel caliber, which represents a moving blood column, whereas the surrounding thin, transparent plasma edge stream cannot be seen. Therefore, photograph-based vessel measurements slightly underestimate true vessel lumen width [20]. Lumen width measurements made on OCT images are also likely to be underestimated, because such measurements do not include the plasma layer against the vessel wall. Vessel OCT measurements rely on outer and inner vessel wall
Blood pressure (mmHg)
Intraocular pressure (mmHg)
13.5 ± 3.1 (8.3–20.0)
Heart rate (beats/min)
71.6 ± 13.5
Axial length (mm)
25.1 ± 0.8
Data presented as mean ± standard deviation, where applicable
p \ 0.0001; Fig. 3). Additionally, ACH and ACV were not significantly different (p = 0.16). However, ALH measurements were significantly higher than ALV measurements (p = 0.03). Mean WTH and WLRH were 21.6 ± 2.87 lm and 0.21 ± 0.03, respectively. Mean WTV and WLRV were 26.5 ± 3.27 lm and 0.27 ± 0.04, respectively. There was a significant correlation between WLRH and WLRV (r = 0.45, p = 0.02) but not between
123
170
S. Arichika et al.
Fig. 3 Correlations between vessel measurements made with different imaging modalities. Vessel measurements (caliber width and lumen width) made with adaptive optics scanning laser
ophthalmoscopy (AOSLO) and optical coherence tomography (OCT) were significantly correlated
hyperreflectivity, where projected OCT light hits perpendicular to the vessel [12]. Cimalla et al. [21] speculate that vessel wall hyperreflectivity results from a combination of shear-induced red blood cell alignment within vessels and vessel geometry. Therefore, when using OCT images to evaluate vessel wall thickness, caution should be taken because the plasma stratum may be unintentionally included in the measurements. Interestingly, we found no correlation between AOSLO and OCT measurements of wall thickness, possibly because the proportion of plasma stratum relative to vessel caliber or lumen is inconsistent. These OCT limitations do not apply to AOSLO because this system is able to directly image the vessel wall. Although caliber width measurements with the two imaging modalities were strongly correlated, AOSLO measurements of lumen width were significantly higher than OCT measurements. This suggests that OCT measurements include the thickness of the plasma stratum. Muraoka et al. [12] successfully made venous wall measurements with OCT. However, the proportion of plasma stratum thickness in these measurements should not be ignored. This is especially true because even high-resolution AOSLO systems with an offset mode [8] do not always reliably image very thin venous walls. Our study had several limitations. First, the exact target areas of AOSLO and OCT measurements were slightly different. All OCT measurements were made 1.73 mm from the center of the optic disc, slightly inside zone B (0.5–1 disc diameters from the optic disc margin), but the IVAN system used AOSLO measurements made 1.80 mm from the optic disc center. It is likely that this slight difference only negligibly influenced our results because the AOSLO imaging area was a 0.34-mm 9 0.34-mm square that included the OCT imaging circle. Second, because the
OCT scan circle crosses various vessels around the disc, OCT measurements were calculated using one point for each vessel. In contrast, AOSLO measurements were the average of multiple measurements taken every 6 lm along the same vessel. Third, OCT images provided vertical vascular information, while AOSLO images provided horizontal vascular information. Lastly, the shape of retinal vessels was assumed to be completely round, but this is not always the case. Muraoka et al. [12] previously reported a significant correlation between lumen width measured with OCT and vessel width measured with fundus photographs. Additionally, we found an ICC of 0.759 for AOSLO and OCT caliber width measurements and an ICC of 0.784 for AOSLO and OCT lumen width measurements. This supports the idea that there is little physiological difference between horizontal and vertical lumen width. As shown in Fig. 3, significant correlations were found between ACH and ACV and between ALH and ALV. However, vessels of one volunteer showed a discrepancy between AOSLO and OCT measurements. This may have occurred if the configuration of the retinal artery was not round. To the best of our knowledge, there are no prior comparisons of horizontal and vertical vessel dimensions in humans, and it may have been that this volunteer had a wider horizontal than vertical caliber and lumen. In conclusion, AOSLO- and OCT-based vessel measurements were well correlated. Moreover, unlike OCT images, AOSLO images show plasma and directly visualize the retinal vessel wall. Thus, AOSLO has the potential to measure vessel width more precisely than OCT.
123
Acknowledgments This work was supported, in part, by the Innovative Techno-Hub for Integrated Medical Bio-Imaging of the Project for Developing Innovation Systems, from the Ministry of Education, Culture, Sports, Science and Technology (MEXT) in Japan.
Comparison of retinal vessel measurements using adaptive optics scanning laser… Conflicts of interest S. Arichika, None; A. Uji, None; S. Ooto, None; Y. Muraoka, None; N. Yoshimura, Financial support (Topcon Corporation, Nidek, Canon), Consultant (Nidek).
References 1. Scoles D, Sulai YN, Langlo CS, Fishman GA, Curcio CA, Carroll J, et al. In vivo imaging of human cone photoreceptor inner segments. Invest Ophthalmol Vis Sci. 2014;55:4244–51. 2. Roorda A, Williams DR. The arrangement of the three cone classes in the living human eye. Nature. 1999;397:520–2. 3. Takayama K, Ooto S, Hangai M, Ueda-Arakawa N, Yoshida S, Akagi T, et al. High-resolution imaging of retinal nerve fiber bundles in glaucoma using adaptive optics scanning laser ophthalmoscopy. Am J Ophthalmol. 2013;155:870–81. 4. Tam J, Tiruveedhula P, Roorda A. Characterization of single-file flow through human retinal parafoveal capillaries using an adaptive optics scanning laser ophthalmoscope. Biomed Opt Express. 2011;2:781–93. 5. Arichika S, Uji A, Murakami T, Unoki N, Yoshitake S, Dodo Y, et al. Retinal hemorheological characterization of early-stage diabetic retinopathy using adaptive optics scanning laser ophthalmoscopy. Invest Ophthalmol Vis Sci. 2014;55:8513–22. 6. Bedggood P, Metha A. Direct visualization and characterization of erythrocyte flow in human retinal capillaries. Biomed Opt Express. 2012;3:3264–77. 7. Schallek J, Geng Y, Nguyen H, Williams DR. Morphology and topography of retinal pericytes in the living mouse retina using in vivo adaptive optics imaging and ex vivo characterization. Invest Ophthalmol Vis Sci. 2013;54:8237–50. 8. Chui TY, Vannasdale DA, Burns SA. The use of forward scatter to improve retinal vascular imaging with an adaptive optics scanning laser ophthalmoscope. Biomed Opt Express. 2012;3:2537–49. 9. Koch E, Rosenbaum D, Brolly A, Sahel JA, Chaumet-Riffaud P, Girerd X, et al. Morphometric analysis of small arteries in the human retina using adaptive optics imaging: relationship with blood pressure and focal vascular changes. J Hypertens. 2014;32:890–8. 10. Harazny JM, Ritt M, Baleanu D, Ott C, Heckmann J, Schlaich MP, et al. Increased wall:lumen ratio of retinal arterioles in male patients with a history of a cerebrovascular event. Hypertension. 2007;50:623–9.
171
11. Rizzoni D, Porteri E, Duse S, De Ciuceis C, Rosei CA, La Boria E, et al. Relationship between media-to-lumen ratio of subcutaneous small arteries and wall-to-lumen ratio of retinal arterioles evaluated noninvasively by scanning laser Doppler flowmetry. J Hypertens. 2012;30:1169–75. 12. Muraoka Y, Tsujikawa A, Kumagai K, Akiba M, Ogino K, Murakami T, et al. Age- and hypertension-dependent changes in retinal vessel diameter and wall thickness: an optical coherence tomography study. Am J Ophthalmol. 2013;156:706–14. 13. Ritt M, Harazny JM, Ott C, Schlaich MP, Schneider MP, Michelson G, et al. Analysis of retinal arteriolar structure in never-treated patients with essential hypertension. J Hypertens. 2008;26:1427–34. 14. Cuspidi C, Sala C. Retinal wall-to-lumen ratio: a new marker of endothelial function? J Hypertens. 2011;29:33–5. 15. Arichika S, Uji A, Ooto S, Miyamoto K, Yoshimura N. Adaptive optics-assisted identification of preferential erythrocyte aggregate pathways in the human retinal microvasculature. PLoS ONE. 2014;9:e89679. 16. Bennett AG, Rudnicka AR, Edgar DF. Improvements on Littmann’s method of determining the size of retinal features by fundus photography. Graefes Arch Clin Exp Ophthalmol. 1994;232:361–7. 17. Arichika S, Uji A, Hangai M, Ooto S, Yoshimura N. Noninvasive and direct monitoring of erythrocyte aggregates in human retinal microvasculature using adaptive optics scanning laser ophthalmoscopy. Invest Ophthalmol Vis Sci. 2013;54:4394–402. 18. Uji A, Hangai M, Ooto S, Takayama K, Arakawa N, Imamura H, et al. The source of moving particles in parafoveal capillaries detected by adaptive optics scanning laser ophthalmoscopy. Invest Ophthalmol Vis Sci. 2012;53:171–8. 19. Hubbard LD, Brothers RJ, King WN, Clegg LX, Klein R, Cooper LS, et al. Methods for evaluation of retinal microvascular abnormalities associated with hypertension/sclerosis in the Atherosclerosis Risk in Communities Study. Ophthalmology. 1999;106:2269–80. 20. Baleanu D, Ritt M, Harazny J, Heckmann J, Schmieder RE, Michelson G. Wall-to-lumen ratio of retinal arterioles and arterioleto-venule ratio of retinal vessels in patients with cerebrovascular damage. Invest Ophthalmol Vis Sci. 2009;50:4351–9. 21. Cimalla P, Walther J, Mittasch M, Koch E. Shear flow-induced optical inhomogeneity of blood assessed in vivo and in vitro by spectral domain optical coherence tomography in the 1.3 lm wavelength range. J Biomed Opt. 2011;16:116020.
123
