Developing quantitative analysis program of blood flow velocity according to vessel diameter for neovascular age-related macular degeneration using OCTA-VISTA.

This study aimed to develop a quantitative analysis program of blood flow velocity by vessel diameter in neovascular age-related macular degeneration (nAMD) subjects using high-speed swept-source optical coherence tomography angiography. This retrospective, observational, cross-sectional study included 10 eyes of healthy volunteers and 4 eyes of patients with representative nAMD. Novel scan patterns and variable interscan time analysis were utilized to measure the flow parameter, a surrogate marker of blood flow velocity, by vessel diameter within different depths. Detected vessels at superficial and deep as well as outer retinal regions were categorized into three vessel diameters (major vessels (> 40 µm), medium vessels (20-40 µm), and capillaries (< 20 µm)). The flow parameter increased with enlarged vessel diameter in all participants at superficial and deep layer. All nAMD subjects, except for type 3 macular neovascularization (MNV), contained a structure dominated by medium vessels at outer retinal region. The mean flow parameter at outer retinal region was type 1 MNV (1.46 ms-1), type 1 + 2 MNV (0.98 ms-1), and polypoidal choroidal vasculopathy, including branching vascular networks (1.46 ms-1). This program provides the possibility to extract the blood flow information at different depths by vessel diameter types, which is considered to be useful tool for evaluating nAMD pathology and activity.

Extended and adjustable field-of-view of variable interscan time analysis by ammonite-scanning swept-source optical coherence tomography angiography.

A novel scanning protocol, ammonite scan, is proposed for widefield optical coherence tomography angiography (OCTA) and relative retinal blood flow velocity imaging in the human retina using variable interscan time analysis (VISTA). A repeated circle scan using a 400 kHz swept-source was employed to achieve an interscan time of 1.28 ms. The center of the repeated circular scan continuously moved spirally towards the peripheral region, ensuring an extended and adjustable scan range while preserving the short interscan time. Image artifacts due to eye movement were eliminated via extra motion-correction processing using data redundancy. The relative blood flow velocity in superficial and deep plexus layers was calculated from the VISTA image, and their ratio was used to explore the microvascular flow parameter in the healthy human eye.

EVALUATION OF MORNING GLORY SYNDROME BY SWEPT SOURCE OPTICAL COHERENCE TOMOGRAPHY.

PURPOSE: The purpose of this study was to investigate the deep structural abnormalities in the patients with morning glory syndrome by using swept source optical coherence tomography. METHODS: The papillary and peripapillary areas of the four eyes (four patients) with morning glory syndrome were examined with a prototype swept source optical coherence tomography system with a center wavelength of 1,050 nm. In particular, the abnormalities of the lamina cribrosa and peripapillary sclera were evaluated. RESULTS: The lamina cribrosa was posteriorly displaced in all the four eyes, and the disk area was filled with herniated retinal tissue. A bump-like protrusion near the papillary margin or within the cup area was observed in three eyes. Some tissues were observed to exist between the retina and the sclera surrounding the papillary margin annularly. In one case, swept source optical coherence tomography detected intrascleral adipose tissue in a relatively wide area around the fovea and the optic nerve. CONCLUSION: Swept source optical coherence tomography clearly delineated the papillary and peripapillary abnormalities of morning glory syndrome. Scleral structural abnormalities were present outside the optic nerve as well.

Glaucoma Diagnostic Performance of Retinal Blood Flow Measurement With Doppler Optical Coherence Tomography.

Purpose: The purpose of this study was to evaluate the diagnostic performance of retinal blood flow (RBF) measured with the Doppler optical coherence tomography (OCT) segmental scanning method to distinguish between healthy and glaucoma eyes. Methods: Fifty-eight patients with normal tension glaucoma (NTG) who had a single-hemifield visual field defect and 44 age-matched healthy subjects were enrolled. Retinal nerve fiber layer thickness (RNFLT) was measured with swept-source OCT. Superior and inferior temporal arteries (TAs) and temporal veins (TVs) RBF were measured with Doppler OCT. The area under the curve (AUC) of the receiver operating characteristic (ROC) was used to compare the diagnostic performances in the damaged and normal hemispheres. Results: Multivariate regression analysis showed TA RBF and TV RBF were significantly reduced in the damaged and normal hemispheres. The ROC analysis showed that the AUC for quadrant RNFLT, TA RBF, and TV RBF were 0.973, 0.909, and 0.872 in the damaged hemisphere, respectively. The AUC values in the normal hemisphere were 0.783, 0.744, and 0.697, respectively. The combination of quadrant RNFLT and TA/TV RBF had a greater AUC than quadrant RNFLT alone in both damaged (AUC = 0.987) and normal (AUC = 0.825) hemispheres. Conclusions: In NTG eyes with single-hemifield damage, the RBF was found to be significantly reduced in the damaged and normal hemispheres independent from structural changes. The combination of RNFLT and RBF could improve diagnostic performances for glaucoma. Translational Relevance: Combining morphological and blood flow measurements with Doppler OCT may be useful in glaucoma diagnosis.

Clinical characteristics of glaucoma patients with various risk factors.

BACKGROUND: Glaucoma is multifactorial, but the interrelationship between risk factors and structural changes remains unclear. Here, we adjusted for confounding factors in glaucoma patients with differing risk factors, and compared differences in structure and susceptible areas in the optic disc and macula. METHODS: In 458 eyes with glaucoma, we determined confounding factors for intraocular pressure (IOP), central corneal thickness (CCT), axial length (AL), LSFG-measured ocular blood flow (OBF), which was assessed with laser speckle flowgraphy-measured mean blur rate in the tissue area (MT) of the optic nerve head, biological antioxidant potential (BAP), and systemic abnormalities in diastolic blood pressure (dBP). To compensate for measurement bias, we also analyzed corrected IOP (cIOP; corrected for CCT) and corrected MT (cMT; corrected for age, weighted retinal ganglion cell count, and AL). Then, we determined the distribution of these parameters in low-, middle-, and high-value subgroups and compared them with the Kruskal-Wallis test. Pairwise comparisons used the Steel-Dwass test. RESULTS: The high-cIOP subgroup had significantly worse mean deviation (MD), temporal, superior, and inferior loss of circumpapillary retinal nerve fiber layer thickness (cpRNFLT), and large cupping. The low-CCT subgroup had temporal cpRNFLT loss; the high-CCT subgroup had low cup volume. The high-AL subgroup had macular ganglion cell complex thickness (GCCT) loss; the low-AL subgroup had temporal cpRNFLT loss. The high-systemic-dBP subgroup had worse MD, total, superior, and inferior cpRNFLT loss and macular GCCT loss. The low-BAP subgroup had more male patients, higher dBP, and cpRNFLT loss in the 10 o'clock area. The high-OBF subgroup had higher total, superior and temporal cpRNFLT and macular GCCT. CONCLUSIONS: Structural changes and local susceptibility to glaucomatous damage show unique variations in patients with different risk factors, which might suggest that specific risk factors induce specific types of pathogenesis and corresponding glaucoma phenotypes. Our study may open new avenues for the development of precision medicine for glaucoma.

Correlation Between Enlargement of Retinal Nerve Fiber Defect Angle in En Face Imaging and Visual Field Progression.

Purpose: Retinal nerve fiber layer defects (RNFLDs) become enlarged with glaucoma progression. We measured the RNFLD angle and investigated whether it was correlated with deterioration of the visual field in patients with glaucoma. Methods: This study included 84 eyes of 84 patients with open-angle glaucoma (mean deviation [MD] = -6.51 ± 5.91 dB, follow-up period = 2.82 ± 0.74 years) with the RNFLDs, who underwent en face swept-source optical coherence tomography (SS-OCT) wide scans (12 × 9 mm) at least 6 times. The RNFLD angle was measured as the intersection between the RNFLD and a circle centered on the disc with a radius half the distance between the disc and the fovea. Slopes for the RNFLD angle, macular ganglion cell layer thickness (GCCT), and circumpapillary RNFL thickness (cpRNFLT) were compared with the MD slope, as measured with the Humphrey field analyzer 24-2 program. Results: The correlation coefficients with MD slope were -0.67 for the RNFLD angle slope (P < 0.001), 0.15 for the macular GCCT slope (P = 0.163), and 0.04 for the cpRNFLT slope (P = 0.719). The RNFLD angle tended to increase as the number of disc hemorrhage occurrences increased (rs = 0.31, P = 0.004). The RNFLD angle slope also had good predictive power for glaucoma progression (area under the receiver operating characteristic curve = 0.88, 95% confidence interval = 0.81-0.95). Conclusions: We found that the RNFLD angle slope was more closely associated with the MD slope than were other OCT parameters. This suggests that measurement of the RNFLD angle with en face OCT images could be effective in evaluating glaucoma progression. Translational Relevance: Our study provides a method for monitoring glaucoma progression with SS-OCT.

Novel volumetric imaging biomarkers for assessing disease activity in eyes with PCV.

The aim of this study was to evaluate influence of baseline imaging features on visual and anatomical outcomes in eyes with PCV treated with anti-VEGF monotherapy. In this prospective study we enrolled participants with treatment-naïve PCV who followed a treat-and-extend protocol using intravitreal aflibercept (IVA) monotherapy. Baseline clinical features evaluatedincluded best corrected visual acuity (BCVA), traditional features such as lesion size, fluid-related OCT parameters and novel parameters using automated software. This included quantitative and qualitative pigment epithelium detachment (PED) parameters [height, volume]; and choroidal parameters. [choroidal thickness (CT), choroidal volume (CV) and choroidal vascularity index (CVI). We evaluated the predictive value of each parameter on visual and anatomical outcome at month 12. We additionally evaluated initial treatment response after 3 monthly injections with respect to month 12 outcomes. Fifty-two eyes from 52 participants were included in the study. The BCVA increased from 61.1 ± 13.2 to 69.6 ± 13.2 early treatment diabetic retinopathy study (ETDRS) letters (p < 0.01) and CRT reduced from 455.7 ± 182.4 µm to 272.7 ± 86.2 (p < 0.01) from baseline to month 12. The proportion of eyes with PED decreased significant from 100% at baseline to 80% at month 12 (p < 0.01). Reduction in the mean maximum height of PED (from 381.3 ± 236.3 µm to 206.8 vs ± 146.4 µm) and PED volume (from 1322 ± 853 nl to 686 ± 593 nl) (p < 0.01) was also noted from baseline to month12. Baseline features associated with better month 12 BCVA included baseline BCVA (ß = - 0.98, 95%CI - 3.38 to - 1.61, p = 0.02) and baseline CRT (ß = - 0.98, 95%CI - 1.56 to - 0.40, p = 0.04) while the disease activity at month12 was significantly associated with lower baseline CRT (366.0 ± 129.5 vs 612.0 ± 188.0 , p < 0.001), lower baseline PED height (242.0 ± 150.0 vs 542.0 ± 298.0 µm, p < 0.01), lower baseline PED volume (0.6 ± 0.3 mm3 vs 2.2 ± 1.3 mm3 vs, p < 0.01), lower proportion with marked CVH (17.9% vs 46.2%, p = 0.02) and lower mean CVI (61.8 ± 1.4 vs 63.0 ± 1.4, p < 0.02). Additionally, a larger decrease in CRT (per 100 nm) and larger PED volume reduction (per 100 nl) at month 3 from baseline were associated with greater BCVA gain and inactive disease. PED-related volumetric parameters have an additional predictive value to traditional biomarkers of disease activity in eyes with PCV undergoing anti-VEGF monotherapy. With increasingly precise quantification, PEDs can be a crucial biomarker in addition to traditional parameters and may aid in retreatment decisions.

Sensorless astigmatism correction using a variable cross-cylinder for high lateral resolution optical coherence tomography in a human retina.

High lateral resolution (∼5µm) optical coherence tomography (OCT) that employs a variable cross-cylinder (VCC) to compensate for astigmatism is presented for visualizing minute structures of the human retina. The VCC and its sensorless optimization process enable ocular astigmatism correction of up to -5.0 diopter within a few seconds. VCC correction has been proven to increase the signal-to-noise ratio and lateral resolution using a model eye. This process is also validated using the human eye by visualizing the capillary network and human cone mosaic. The proposed method is applicable to existing OCT, making high lateral resolution OCT practical in clinical settings.

Retinal Blood Velocity Waveform Characteristics With Aging and Arterial Stiffening in Hypertensive and Normotensive Subjects.

Purpose: We aimed to explore the velocity waveform characteristics of the retinal artery associated with age and the cardio-ankle vascular index (CAVI) as a conventional arterial stiffness marker by applying the Doppler optical coherence tomography (DOCT) flowmeter. Methods: In this cross-sectional study, DOCT flowmeter imaging was performed in 66 participants aged 21 to 83 years (17 men, 49 women) with no history of eye diseases and no systemic diseases, except for hypertension. Retinal blood velocity waveform was analyzed where several parameters in time (upstroke time, T1, T2, T3, and T4) and area under the waveform (area elevation, area declination, A1, A2, A3, and A4) were extracted. Systolic blood pressure-adjusted Pearson's coefficients were calculated to determine the correlations of each parameter with age or CAVI. Results: Corrected upstroke time (UTc) was the waveform parameter most positively correlated with age (r = 0.497, P < 0.001). Area declination was the waveform parameter most negatively correlated with age (r = -0.682, P < 0.001) and CAVI (r = -0.601, P < 0.001). Conclusions: We extracted the waveform parameters associated with the risks of arterial stiffening. The velocity waveform analysis of the retinal artery with DOCT flowmeter potentially could become a new method for arterial stiffness identification. Translational Relevance: DOCT flowmeter could evaluate arterial stiffening in a different way from the conventional method of measuring arterial stiffening using pressure waveform. Because the DOCT flowmeter can easily, quickly, and noninvasively provide a retinal blood velocity waveform, this system could be useful as a routine medical examination for arterial stiffening.

Artificial intelligence for classifying uncertain images by humans in determining choroidal vascular running pattern and comparisons with automated classification between artificial intelligence.

PURPOSE: Abnormalities of the running pattern of choroidal vessel have been reported in eyes with pachychoroid diseases. However, it is difficult for clinicians to judge the running pattern with high reproducibility. Thus, the purpose of this study was to compare the degree of concordance of the running pattern of the choroidal vessels between that determined by artificial intelligence (AI) to that determined by experienced clinicians. METHODS: The running pattern of the choroidal vessels in en face images of Haller's layer of 413 normal and pachychoroid diseased eyes was classified as symmetrical or asymmetrical by human raters and by three supervised machine learning models; the support vector machine (SVM), Xception, and random forest models. The data from the human raters were used as the supervised data. The accuracy rates of the human raters and the certainty of AI's answers were compared using confidence scores (CSs). RESULTS: The choroidal vascular running pattern could be determined by each AI model with an area under the curve better than 0.94. The random forest method was able to discriminate with the highest accuracy among the three AIs. In the CS analyses, the percentage of certainty was highest (66.4%) and that of uncertainty was lowest (6.1%) in the agreement group. On the other hand, the rate of uncertainty was highest (27.3%) in the disagreement group. CONCLUSION: AI algorithm can automatically classify with ambiguous criteria the presence or absence of a symmetrical blood vessel running pattern of the choroid. The classification was as good as that of supervised humans in accuracy and reproducibility.

Association between Topographic Features of the Retinal Nerve Fiber Bundle and Good Visual Acuity in Patients with Glaucoma.

Purpose: Maintaining visual acuity in glaucoma patients is an important part of preventing the deterioration of quality of vision. We identified specific areas of the papillomacular bundle (PMB) that were strongly associated with visual acuity, based on en-face images derived from optical coherence topography (OCT) wide scans.Methods: The study recruited 23 eyes of 21 glaucoma patients (age: 61.3 ± 13.0 years, M: F = 9:12, Humphrey field analyzer-measured mean deviation: -19.9 ± 6.5 dB) with good best-corrected visual acuity (20/20 or more) and a remaining PMB with a maximum width no more than half that of the vertical disc diameter. En-face images were derived from 12 × 9 mm wide-scan images made with DRI-OCT (Triton, Topcon). Averaged en-face images were created by identifying the disc center and fovea line (DFL) and aligning it between images. We then measured the frequency of remaining PMB at 10 µm intervals along a vertical line intersecting the DFL at its midpoint. Finally, we used a logistic analysis in a much larger group of patients to identify cases of glaucoma with low BCVA (<20/20).Results: In the averaged en-face image, the residual PMB area appeared as a high-intensity region above the DFL. Analysis showed that residual PMB was most common in an area 830-870 µm above the DFL. The correlation coefficient of residual PMB in this area to BCVA was -0.57 (p < .01), and among OCT parameters in this residual PMB area, the AUC to identify decreased BCVA was highest for ganglion cell complex thickness (0.85, p < .01), with a cutoff of 87.5 µm.Conclusions: This study identified specific areas of the PMB that were associated with BCVA in wide-scan, en-face OCT images from glaucoma patients. This suggests that it may be possible to identify visual impairment during glaucoma treatment by measuring this area.

Hierarchical deep learning models using transfer learning for disease detection and classification based on small number of medical images.

Deep learning is being employed in disease detection and classification based on medical images for clinical decision making. It typically requires large amounts of labelled data; however, the sample size of such medical image datasets is generally small. This study proposes a novel training framework for building deep learning models of disease detection and classification with small datasets. Our approach is based on a hierarchical classification method where the healthy/disease information from the first model is effectively utilized to build subsequent models for classifying the disease into its sub-types via a transfer learning method. To improve accuracy, multiple input datasets were used, and a stacking ensembled method was employed for final classification. To demonstrate the method's performance, a labelled dataset extracted from volumetric ophthalmic optical coherence tomography data for 156 healthy and 798 glaucoma eyes was used, in which glaucoma eyes were further labelled into four sub-types. The average weighted accuracy and Cohen's kappa for three randomized test datasets were 0.839 and 0.809, respectively. Our approach outperformed the flat classification method by 9.7% using smaller training datasets. The results suggest that the framework can perform accurate classification with a small number of medical images.

CS2-Net: Deep learning segmentation of curvilinear structures in medical imaging.

Automated detection of curvilinear structures, e.g., blood vessels or nerve fibres, from medical and biomedical images is a crucial early step in automatic image interpretation associated to the management of many diseases. Precise measurement of the morphological changes of these curvilinear organ structures informs clinicians for understanding the mechanism, diagnosis, and treatment of e.g. cardiovascular, kidney, eye, lung, and neurological conditions. In this work, we propose a generic and unified convolution neural network for the segmentation of curvilinear structures and illustrate in several 2D/3D medical imaging modalities. We introduce a new curvilinear structure segmentation network (CS2-Net), which includes a self-attention mechanism in the encoder and decoder to learn rich hierarchical representations of curvilinear structures. Two types of attention modules - spatial attention and channel attention - are utilized to enhance the inter-class discrimination and intra-class responsiveness, to further integrate local features with their global dependencies and normalization, adaptively. Furthermore, to facilitate the segmentation of curvilinear structures in medical images, we employ a 1×3 and a 3×1 convolutional kernel to capture boundary features. Besides, we extend the 2D attention mechanism to 3D to enhance the network's ability to aggregate depth information across different layers/slices. The proposed curvilinear structure segmentation network is thoroughly validated using both 2D and 3D images across six different imaging modalities. Experimental results across nine datasets show the proposed method generally outperforms other state-of-the-art algorithms in various metrics.

Effect of peripapillary tilt direction and magnitude on central visual field defects in primary open-angle glaucoma with high myopia.

PURPOSE: To evaluate the relationship between peripapillary tilt and visual field (VF) defects in glaucomatous eyes with axial myopia. STUDY DESIGN: Retrospective cross-sectional study. PATIENTS AND METHODS: One hundred four eyes of 104 patients with primary open-angle glaucoma (POAG) with myopia were included (52 eyes with high myopia [HM], 26.5 mm ≤ axial length [AL] < 30.0 mm; and 52 eyes without HM, 24.0 mm < AL < 26.5 mm). The direction and magnitude of the peripapillary tilt were evaluated using optical coherence tomography. The eyes were divided into 12 groups according to the tilt directions defined by clock-hour sectors in a clockwise direction in the right eyes and in a counterclockwise direction in the left eyes. The mean deviation (MD) and central VF (CVF) values, ie, the mean threshold values of 4 paracentral points within 5 degrees of the Swedish Interactive Threshold Algorithm 30-2 test, were evaluated. RESULTS: The direction of the tilt was toward sector 9 (47.1%) and sector 8 (34.6%). The MD and CVF values were significantly worse (P = 0.013 and P = 0.019, respectively) in the sector 9 group than in the sector 8 group. Furthermore, the smaller peripapillary tilt magnitude in the sector 9 group was negatively correlated (P = 0.0019) with the CVF but not with the MD (P = 0.1) among the POAG eyes with HM. In contrast, the ovality index in the sector 9 group was not significantly correlated with the MD (P = 0.4) or the CVF (P = 0.36). CONCLUSION: A smaller temporal peripapillary tilt correlated with CVF defects in POAG eyes with HM. The peripapillary tilt direction and magnitude affect the CVF defect in POAG eyes with HM.

Structural Characterization of Glaucoma Patients with Low Ocular Blood Flow.

Purpose: There is an unclear relationship between ocular blood flow (OBF) and the structural characteristics of the optic nerve head (ONH) in glaucoma, a multifactorial disease. This study used laser speckle flowgraphy (LSFG) to identify low-OBF glaucoma patients and investigated the ONH in these patients. Materials and Methods: In 533 eyes with glaucoma, we determined confounding factors for LSFG-measured OBF (tissue-area mean blur rate: MT) and corrected MT with a linear mixed-effects model (LMM). Structural ONH data (from fundus stereo photography), OCT data, and clinical characteristics were then compared in patients with corrected MT in the upper and lower quartiles using the LMM. Results: Single regression showed significant correlations between MT and age, spherical equivalent (SE), central corneal thickness (CCT), and a weighted count of retinal ganglion cells (wRGC), but not axial length or systemic blood pressure. Gender also significantly influenced MT; MT was corrected for these correlated factors and also glaucoma type with the LMM. The lower-quartile MT group had a significantly larger cup area and cup-disc area ratio and lower temporal quadrant circumpapillary retinal nerve fiber layer thickness (cpRNFLT) and macular ganglion cell complex (GCC) than the upper-quartile group. Conclusions: Low-OBF glaucoma patients were characterized by a larger cup-disc ratio, and higher susceptibility to damage in the temporal disc and the macular area.

Optical Coherence Tomography-Based Deep-Learning Models for Classifying Normal and Age-Related Macular Degeneration and Exudative and Non-Exudative Age-Related Macular Degeneration Changes.

INTRODUCTION: The use of optical coherence tomography (OCT) images is increasing in the medical treatment of age-related macular degeneration (AMD), and thus, the amount of data requiring analysis is increasing. Advances in machine-learning techniques may facilitate processing of large amounts of medical image data. Among deep-learning methods, convolution neural networks (CNNs) show superior image recognition ability. This study aimed to build deep-learning models that could distinguish AMD from healthy OCT scans and to distinguish AMD with and without exudative changes without using a segmentation algorithm. METHODS: This was a cross-sectional observational clinical study. A total of 1621 spectral domain (SD)-OCT images of patients with AMD and a healthy control group were studied. The first CNN model was trained and validated using 1382 AMD images and 239 normal images. The second transfer-learning model was trained and validated with 721 AMD images with exudative changes and 661 AMD images without any exudate. The attention area of the CNN was described as a heat map by class activation mapping (CAM). In the second model, which classified images into AMD with or without exudative changes, we compared the learning stabilization of models using or not using transfer learning. RESULTS: Using the first CNN model, we could classify AMD and normal OCT images with 100% sensitivity, 91.8% specificity, and 99.0% accuracy. In the second, transfer-learning model, we could classify AMD as having or not having exudative changes, with 98.4% sensitivity, 88.3% specificity, and 93.9% accuracy. CAM successfully described the heat-map area on the OCT images. Including the transfer-learning model in the second model resulted in faster stabilization than when the transfer-learning model was not included. CONCLUSION: Two computational deep-learning models were developed and evaluated here; both models showed good performance. Automation of the interpretation process by using deep-learning models can save time and improve efficiency. TRIAL REGISTRATION: No15073.

Retinal Arteriole Pulse Waveform Analysis Using a Fully-Automated Doppler Optical Coherence Tomography Flowmeter: a Pilot Study.

PURPOSE: To evaluate the repeatability and reproducibility of the measurement of retinal arteriole pulse waveforms using a novel fully-automated Doppler optical coherence tomography (DOCT) flowmeter in healthy subjects. METHODS: Twenty eyes of 20 healthy subjects were included to test the intrasession repeatability of pulse waveform analysis. DOCT measurements were performed based on a newly developed instantaneous Doppler angle measurement method. Upstroke time (UT), which is the time from the minimum to the maximum retinal blood velocity, and the resistance index (RI) of the retinal arteriole pulse waveform were measured. Coefficients of variation (CVs) and intraclass correlation coefficients (ICCs) were calculated. Interdevice reproducibility between two instruments was assessed in five eyes of five subjects. RESULTS: The mean UT was 130.3 ms (range, 110.1-152.1 ms), and the mean RI was 0.66 (range, 0.51-0.82). The respective ICCs of UT and the RI for the intrasession repeatability of assessment were 0.87 and 0.78. The respective CVs of UT and the RI were 6.6 ± 3.3% and 4.7 ± 2.1%. Regarding interdevice reproducibility, there were no significant differences between the measurements derived from the instruments (P > 0.05). CONCLUSIONS: Pulse waveform measurement in retinal arterioles using a fully-automated DOCT flowmeter exhibited good repeatability and interdevice reproducibility. TRANSLATIONAL RELEVANCE: The above-described improved DOCT flowmeter system provides reasonably repeatable measurements of retinal arteriole pulse waveforms, potentially facilitating systemic-circulation abnormality monitoring. The examination of the circulation with the novel device can be potentially useful for evaluating systemic circulation.

Assessment of the deformation of the outer nuclear layer in the Epiretinal membrane using spectral-domain optical coherence tomography.

BACKGROUND: We aimed to investigate the deformation of the outer nuclear layer using optical coherence tomography in patients with epiretinal membrane (ERM) and its relationship with metamorphopsia. METHODS: Thirty-nine eyes from 39 patients with ERM were included in the study. Patients with the subtypes of pseudo macula hole and lamellar hole were excluded. Twenty-one fellow eyes without macular disease were included as normal controls. Forty-nine B-scan images were obtained in the range of 20 degrees around the macula using SD-OCT. The outer nuclear layer (ONL) was evaluated as a three-dimensional image (3D-ONL) reconstructed using the distance between the ONL and retinal pigment epithelium (RPE) line. The deformation of the ONL was figured at the reference plane and evaluation plane (ONL-B). The characteristic parameters of the ONL-B were defined as circularity, area ratio, and axis ratio. The correlations between these parameters and visual acuity and MCHART score ratio (MH/MV) were then evaluated. RESULTS: ONL height was significantly higher in ERM patients than in normal controls (54.1 ± 5.3 µm and 84.1 ± 12.9 µm, respectively; P < 0.001). In ERM patients, the MV score was 0.53 ± 0.50, the MH score was 0.71 ± 0.61, and the distance from the RPE line to the ONL-B was 153.5 ± 13.5 µm. The axis of the ONL-B in normal controls and ERM patients was - 6.25 ± 21.8 and - 1.28 ± 29.1, respectively, which indicates that the ONL is horizontally long in both normal individuals and ERM patients. The circularity and area ratio were significantly smaller in ERM patients than in normal controls. In all ERM patients, MH/MV had a significant correlation with axis (r = - 0.29, p = 0.034), circularity (r = - 0.28, p = 0.044), and area ratio (r = - 0.47, p = 0.001). Moreover, we found that the correlation was more significant if the subjects had an axis of the ONL within ±10 degrees (n = 16); the correlations of MH/MV with axis (r = - 0.29, p = 0.034), circularity (r = - 0.53, p = 0.021), and area ratio were more significant (r = - 0.78, P < 0.0001). CONCLUSION: The ONL is horizontally long in normal individuals and ERM patients. The direction of metamorphopsia is correlated with the direction of ONL deformation.

Glaucoma Diagnosis with Machine Learning Based on Optical Coherence Tomography and Color Fundus Images.

This study aimed to develop a machine learning-based algorithm for glaucoma diagnosis in patients with open-angle glaucoma, based on three-dimensional optical coherence tomography (OCT) data and color fundus images. In this study, 208 glaucomatous and 149 healthy eyes were enrolled, and color fundus images and volumetric OCT data from the optic disc and macular area of these eyes were captured with a spectral-domain OCT (3D OCT-2000, Topcon). Thickness and deviation maps were created with a segmentation algorithm. Transfer learning of convolutional neural network (CNN) was used with the following types of input images: (1) fundus image of optic disc in grayscale format, (2) disc retinal nerve fiber layer (RNFL) thickness map, (3) macular ganglion cell complex (GCC) thickness map, (4) disc RNFL deviation map, and (5) macular GCC deviation map. Data augmentation and dropout were performed to train the CNN. For combining the results from each CNN model, a random forest (RF) was trained to classify the disc fundus images of healthy and glaucomatous eyes using feature vector representation of each input image, removing the second fully connected layer. The area under receiver operating characteristic curve (AUC) of a 10-fold cross validation (CV) was used to evaluate the models. The 10-fold CV AUCs of the CNNs were 0.940 for color fundus images, 0.942 for RNFL thickness maps, 0.944 for macular GCC thickness maps, 0.949 for disc RNFL deviation maps, and 0.952 for macular GCC deviation maps. The RF combining the five separate CNN models improved the 10-fold CV AUC to 0.963. Therefore, the machine learning system described here can accurately differentiate between healthy and glaucomatous subjects based on their extracted images from OCT data and color fundus images. This system should help to improve the diagnostic accuracy in glaucoma.

Deep Learning Classification Models Built with Two-step Transfer Learning for Age Related Macular Degeneration Diagnosis.

The objective of this study was to build deep learning models with optical coherence tomography (OCT) images to classify normal and age related macular degeneration (AMD), AMD with fluid, and AMD without any fluid. In this study, 185 normal OCT images from 49 normal subjects, 535 OCT images of AMD with fluid, and 514 OCT mages of AMD without fluid from 120 AMD eyes as training data, while 49 normal images from 25 normal eyes, 188 AMD OCT images with fluid and 154 AMD images without any fluid from 77 AMD eyes as test data, were enrolled. Data augmentation was applied to increase the number of images to build deep learning models. Totally, two classification models were built in two steps. In the first step, a VGG16 model pre-trained on ImageNet dataset was transfer learned to classify normal and AMD, including AMD with fluid and/or without any fluid. Then, in the second step, the fine-tuned model in the first step was transfer learned again to distinguish the images of AMD with fluid from the ones without any fluid. With the first model, normal and AMD OCT images were classified with 0.999 area under receiver operating characteristic curve (AUC), and 99.2% accuracy. With the second model, AMD with the presence of any fluid, and AMD without fluid were classified with 0.992 AUC, and 95.1% accuracy. Compared with a transfer learned VGG16 model pre-trained on ImageNet dataset, to classify the three categories directly, higher classification performance was achieved with our notable approach. Conclusively, two classification models for AMD clinical practice were built with high classification performance, and these models should help improve the early diagnosis and treatment for AMD.