Faculty Bibliography

PURPOSE/OBJECTIVE:To determine which spectral domain optical coherence tomography biomarkers of idiopathic macular hole (MH) correlate with the postoperative best-corrected visual acuity (BCVA) in anatomically closed MH. METHODS:Retrospective analysis of spectral domain optical coherence tomography scans of 44 patients presenting with MH followed for a mean of 17 months. Widths of MH aperture, base, and ellipsoid zone disruption were calculated from presenting foveal spectral domain optical coherence tomography B-scans. Macular hole base area and ellipsoid zone disruption area were calculated through the custom in-house software. RESULTS:Poorer postoperative BCVA correlated with increased preoperative choroidal hypertransmission (r = 0.503, P = 0.0005), minimum diameter (r = 0.491, P = 0.0007), and base diameter (r = 0.319, P = 0.0348), but not with preoperative ellipsoid zone width (r = 0.199, P = 0.2001). Applying en-face analysis, the BCVA correlated weakly with preoperative ellipsoid zone loss area (r = 0.380, P = 0.013), but not with preoperative MH base area (r = 0.253, P = 0.1058). CONCLUSION/CONCLUSIONS:Increased MH minimum diameter, base diameter, base area, and choroidal hypertransmission are correlated with a poorer postoperative BCVA. Ellipsoid zone loss measurements were not consistently correlated with a BCVA. Choroidal hypertransmission width may be an easy-to-visualize predictive imaging biomarker in MH surgery.

AIMS/OBJECTIVE:To evaluate the associations between retinal blood flow (RBF) and optical coherence tomography (OCT) structural measurements in normal-tension glaucoma (NTG) eyes with single-hemifield visual field (VF) damage by the Doppler OCT. METHODS:The Doppler OCT was used to measure temporal artery (TA) RBF and temporal vein (TV) RBF. Retinal nerve fibre layer thickness (RNFLT) was measured by spectral-domain OCT. RESULTS:Forty-three consecutive eyes of 43 patients with NTG with VF defect confined to a single hemifield and 24 eyes of 24 age-matched healthy subjects were studied. TA and TV RBF and RNFLT were reduced in the damaged hemisphere compared with the normal hemisphere (mean (SD), 3.61 (1.68) vs 5.86 (2.59) µL/min, p<0.001; 5.61 (2.51) vs 6.94 (2.83) µL/min, p=0.010; 69.0 (19.7) vs 99.7 (22.8) µm, p<0.001). Those values in the normal hemisphere of NTG eyes also decreased compared with the healthy hemisphere of the healthy eyes (8.40 (3.36) µL/min, p<0.001; 9.28 (4.47) µL/min, p<0.002; 122.8 (20.2) µm, p<0.001). Multivariate model showed that normal and damaged hemispheres and RNFLT were associated with RBF reduction. In addition, the RBF in the normal hemisphere was lower than that in the healthy hemisphere even after adjusting for RNFLT. CONCLUSION/CONCLUSIONS:In NTG eyes with single-hemifield damage, the RBF was significantly reduced in the damaged hemisphere compared with the normal one. The RBF decreased in the normal and damaged hemispheres of NTG eyes compared with the healthy hemisphere independent from RNFLT.

Glaucoma is a neurodegenerative disease of the visual system and is the leading cause of irreversible blindness worldwide. To date, its pathophysiological mechanisms remain unclear. This study evaluated the feasibility of advanced diffusion magnetic resonance imaging techniques for examining the microstructural environment of the visual pathway in glaucoma. While conventional diffusion tensor imaging (DTI) showed lower fractional anisotropy and higher directional diffusivities in the optic tracts of glaucoma patients than healthy controls, diffusion kurtosis imaging (DKI) and the extended white matter tract integrity (WMTI) model indicated lower radial kurtosis, higher axial and radial diffusivities in the extra-axonal space, lower axonal water fraction, and lower tortuosity in the same regions in glaucoma patients. These findings suggest glial involvements apart from compromised axonal integrity in glaucoma. In addition, DKI and WMTI but not DTI parameters significantly correlated with clinical ophthalmic measures via optical coherence tomography and visual field perimetry testing. Taken together, DKI and WMTI provided sensitive and comprehensive imaging biomarkers for quantifying glaucomatous damage in the white matter tract across clinical severity complementary to DTI.

Purpose: Rigorous clinical testing has established that Schlemm"™s canal cross-sec-tional area (SC-CSA) is reduced in glaucomatous eyes. However, to date, it is unclear whether trabecular bypass procedures impact the morphology of the proximal aqueous outflow tract, or if the introduction of a local region of low outflow resistance adversely affects SC-CSA elsewhere, specifically presenting as SC diminution. This study quantifies changes in the morphology of the distal outflow pathway after iStent Trabecular Micro-Bypass stent (Glaukos Corp, Laguna Hills, CA, USA) implantation in living eyes by anterior segment optical coherence tomography (OCT). Design: This was a prospective observational study. Subjects: This study included six patients (eight eyes) with primary-open angle glaucoma. Methods: Patients underwent iStent placement in the nasal anterior chamber angle quadrant. OCT imaging was obtained of both nasal and temporal eye quadrants before and after surgery. For each SC parameter, an average of ten consecutive, evenly spaced measurements were manually obtained over a 1 mm segment of SC on FIJI ImageJ. Linear mixed effects modeling quantified the effect of the iStent on these parameters. Main outcome measures: Main outcome measures were changes in SC-CSA, inner-to-outer wall distance (IOD), and trabecular meshwork (TM) thickness following iStent placement. Results: Following iStent placement, total SC-CSA increased an average of 1,039.12 µm² (P = 0.05). Individually, there were no significant changes in SC-CSA in the nasal or temporal quadrants. Total SC-IOD and nasal SC-IOD increased an average of 2.35 µm (P = 0.01) and 2.96 µm (P = 0.04), respectively. There were no significant changes in temporal quadrant SC-IOD. There were no significant changes in TM thickness in either quadrant. Conclusions: Implantation of the iStent Trabecular Micro-Bypass stent significantly increases SC-IOD in the nasal quadrant at the location of implant, with no evidence of SC diminution in the temporal quadrant. It remains unclear how these observa-tions relate to the surgical efficacy of trabecular bypass procedures.

The direct analysis of 3D Optical Coherence Tomography (OCT) volumes enables deep learning models (DL) to learn spatial structural information and discover new bio-markers that are relevant to glaucoma. Downsampling 3D input volumes is the state-of-art solution to accommodate for the limited number of training volumes as well as the available computing resources. However, this limits the network's ability to learn from small retinal structures in OCT volumes. In this paper, our goal is to improve the performance by providing guidance to DL model during training in order to learn from finer ocular structures in 3D OCT volumes. Therefore, we propose an end-to-end attention guided 3D DL model for glaucoma detection and estimating visual function from retinal structures. The model consists of three pathways with the same network architecture but different inputs. One input is the original 3D-OCT cube and the other two are computed during training guided by the 3D gradient class activation heatmaps. Each pathway outputs the class-label and the whole model is trained concurrently to minimize the sum of losses from three pathways. The final output is obtained by fusing the predictions of the three pathways. Also, to explore the robustness and generalizability of the proposed model, we apply the model on a classification task for glaucoma detection as well as a regression task to estimate visual field index (VFI) (a value between 0 and 100). A 5-fold cross-validation with a total of 3782 and 10,370 OCT scans is used to train and evaluate the classification and regression models, respectively. The glaucoma detection model achieved an area under the curve (AUC) of 93.8% compared with 86.8% for a baseline model without the attention-guided component. The model also outperformed six different feature based machine learning approaches that use scanner computed measurements for training. Further, we also assessed the contribution of different retinal layers that are relevant to glaucoma. The VFI estimation model achieved a Pearson correlation and median absolute error of 0.75 and 3.6%, respectively, for a test set of size 3100 cubes.

Purpose : In mild glaucoma, RNFL thinning and visual field (VF) loss are often localized, but structure-function modeling is impeded by variability due to individual eye anatomy. We perform high-resolution spatial correlations of RNFLT maps for each VF location to identify relevant areas and study further improvements by geometrically aligning RNFLT maps based on artery trajectories. Methods : In 419 SITA Standard 24-2 Humphrey VFs with at most mild glaucoma (mean deviation >=-3dB) with accompanying circumpapillary Cirrus HD-OCT RNFLT maps, we computed pixel-wise correlations (52 VF locations x 40401 pixels). We then performed an alignment operation, ensuring that the two major retinal arteries follow the same lines in all scans. We piecewise linearly approximated the trajectories of the arteries on 4 concentric circles around ONH (Fig. 1a), determined the necessary rotation for each pixel, and morphed the images accordingly (Fig. 1b). Results : For the pre-alignment RNFLT (correlation maps Fig. 2 top) we observed: (1) relatively high correlations (max 0.29); (2) most of the high-correlation regions are highly localized around the median trajectories of the major arteries at most VF locations, possibly due to the stacked character of the fiber bundles close to ONH, which impedes precise spatial mapping to the VF. This observation suggests general retinal vulnerability zones rather than highly VF location-specific areas as assumed by many previous structure-function models. Accordingly, morphing the RNFLT maps by aligning the eyespecific artery locations increased the maximal correlations on 25 of the 52 VF locations (Fig. 2 bottom, marked in green), particularly in nasal and inferior VF, with improvements of up to 0.1 (inferior arcuate region of VF). At many locations, aligned vulnerability areas become substantially more conspicuous (e.g. the location enlarged on the top left) and might have been missed without aligning. Conclusions : High-resolution structure-function correlations reveal retinal vulnerability zones in mild glaucoma. At many VF locations, these zones become better correlated with VF regions when RNFLT maps are aligned along the arteries. Specific attention to RNFL thinning in these zones in glaucoma suspects may improve the detection of initial VF loss glaucoma

Purpose : To develop a novel glaucoma progression model with deep learning methods incorporating four major glaucoma biomarkers: VFI, MD, cRNFL and GCIPL. Methods : 1023 eyes from 596 glaucoma/glaucoma-suspect patients were included from clinic. Two types of deep learning (DL) models were developed using Keras: an artificial neural network (ANN) and a recurrent neural network (RNN). The ANN contained five fullyconnected (FC) layers, with a leaky rectified linear unit activation function and a dropout layer with a rate of 0.2. The RNN contained two long short-term memory layers, followed by a FC layer and a dropout layer with a rate of 0.2. Both models were trained to predict four major clinical biomarkers for glaucoma: visual field index (VFI), mean deviation (MD), circumpapillary retinal nerve fiber layer (cRNFL) thickness, and ganglion cell inner plexiform layer (GCIPL) thickness. The models were trained using the first three visits to predict the fourth one year later. Train/validation/test splits were 65/15/20. A linear regression (LR) model was trained and evaluated on the same data for baseline comparison. Evaluation of the actual and predicted values were measured by mean absolute error (MAE). Statistical testing of each biomarker was performed between the DL models and LR model by paired Wilcoxon rank sum test. Results : The mean patient age was 62.4 +/- 12.9 years. The baseline mean cRNFL: 76.9 +/- 13.4 mum, GCIPL: 70.3 +/- 9.9 mum, VFI: 90.3 +/- 17.8%, and MD: -3.76 +/- 6.13 dB. Table shows the MAE between the actual and predicted values of each of the four biomarkers across all three models. The ANN and RNN models showed statistically significantly smaller MAE compared to the LR model. In particular, the ANN model had the lowest MAE and was able to predict all four biomarkers significantly better than the LR model. Conclusions : By harnessing the power of deep learning, we were able to accurately predict future values of both structural and functional measures of glaucomatous change one year later. This is possible as neural networks are able to recognize the intricate interplay between structural and functional changes in glaucoma that otherwise cannot be well captured in a conventional linear regression model

Purpose : Glaucoma is thought to involve neurochemical changes not only in the eye but also the brain's visual system. While excitotoxicity may play a role in glaucoma pathogenesis, it remains controversial whether excess glutamate occurs in this process. In the current study, we investigated alterations in the excitatory-inhibitory balance (E/I balance) in the visual cortex of glaucoma patients. In addition, we examined whether the altered neurochemical balance in the visual cortex is associated with projections of basal nucleus of Meynert (BNM), a major source of cortical cholinergic innervation in the basal forebrain. Methods : 10 glaucoma patients with a wide range of disease severity and 4 age-matched healthy subjects underwent 3-Tesla anatomical MRI, resting-state functional MRI (fMRI), and magnetic resonance spectroscopy (MRS). We used MEGA-PRESS and PRESS sequences to measure the levels of gamma-aminobutyric acid (GABA) and combined glutamate and glutamine (GLX), respectively. Both GABA and GLX were obtained from the same single voxel (2.2x2.2x2.2 cm³) placed along the calcarine sulci and fitted by LCModel software. We normalized the amount of GABA and GLX to N-acetyl-aspartate (NAA) values obtained from MEGA-PRESS, following LCModel guidelines. E/I balance was calculated by dividing the amount of GLX by the amount of GABA. The resting-state fMRI data were analyzed by CONN software. Results : Glaucoma patients had 16.51% higher E/I balance in the visual cortex compared to the healthy control group (Figure 1a). This difference in E/I balance was apparently driven by a 16.85% reduction in GABA (Figure 1b) with no apparent difference in glutamate or glutamine levels between groups (Figure 1c). Furthermore, the E/I balance in the visual cortex was correlated with the functional connectivity between BNM and the visual cortex (Figure 2). Conclusions : The current study shows that the visual cortex of glaucoma patients adopts an excitatory-dominant state that is driven by reduced GABA. This imbalance was associated with the functional connectivity between BNM and the visual cortex, suggesting that weaker projection of BNM to the visual cortex may play a role in the neurochemical changes in the visual cortex of glaucoma patients. Taken together, these findings suggest that widespread functional and metabolic alterations are involved in the brain during glaucoma pathogenesis

Purpose : Gradient class activation maps (grad-CAM) generated by convolutional neural networks (CNN) have qualitatively indicated that these networks are able to identify important regions in OCT scans. Here, we quantitatively analyse these regions to improve our understanding of the CNN decision making process when detecting glaucoma in OCT volumes. Methods : A total of 1110 OCT (Cirrus HD-OCT, Zeiss, Dublin, Ca) scans from both eyes of 624 subjects (139 healthy and 485 glaucomatous patients (POAG)). An end-to-end 3D-CNN network was trained directly on 3D-volumes for glaucoma detection. Grad-CAM was implemented to highlight structures in the volumes that the network relied on. Grad-CAM heatmaps were generated for 3 different convolutional layers and quantitatively validated by occluding the regions with the highest grad-CAM weights (12.5% of original input volumes) and then evaluating the performance drop. Further, 8-retinal layers segmentation method was used to compute the average heatmap weights for each segmented layer separately, and used to identify the layers that were deemed as important for the task. Results : The model achieved an AUC of 0.97 for the test set (110 scans). Occlusion resulted in a 40% drop in performance (Fig.1). The RNFL and photoreceptors showed the highest median weights for grad-CAM heatmaps (0.1 and 0.2, respectively). The retinal pigment epithelium (RPE) and photoreceptors showed higher weights in the glaucomatous scans (Fig.2-a). RNFL had wider range of weights in healthy cases versus POAG ones. Analysis of the B-scans showed that central part around the optic disc (# 85-135) had the highest contribution to the network decision and the heatmap weights were much higher in glaucoma cases than healthy ones across all B-scans (Fig.2-b). Conclusions : The occlusion experiment indicates that the regions identified by the grad-CAMs are in fact pertinent to the glaucoma detection task. The increased emphasis on the photoreceptors in the glaucoma cases may be attributed to the atrophy in the superficial layers which in turn increased the brightness of this structure. This technique can be used to identify new biomarkers learned for other ocular diseases