Faculty Bibliography

BACKGROUND AND OBJECTIVES/OBJECTIVE:Recent studies have suggested that inter-eye differences (IEDs) in peripapillary retinal nerve fiber layer (pRNFL) or ganglion cell+inner plexiform (GCIPL) thickness by spectral-domain optical coherence tomography (SD-OCT) may identify people with a history of unilateral optic neuritis (ON). However, this requires further validation. Machine learning classification may be useful for validating thresholds for OCT IEDs and for examining added utility for visual function tests, such as low-contrast letter acuity (LCLA), in the diagnosis of people with multiple sclerosis (PwMS) and for unilateral ON history. METHODS:Participants were from 11 sites within the International Multiple Sclerosis Visual System (IMSVISUAL) consortium. pRNFL and GCIPL thicknesses were measured using SD-OCT. A composite score combining OCT and visual measures was compared individual measurements to determine the best model to distinguish PwMS from controls. These methods were also used to distinguish those with history of ON among PwMS. ROC curve analysis was performed on a training dataset (2/3 of cohort), then applied to a testing dataset (1/3 of cohort). Support vector machine (SVM) analysis was used to assess whether machine learning models improved diagnostic capability of OCT. RESULTS:Among 1,568 PwMS and 552 controls, variable selection models identified GCIPL IED, average GCIPL thickness (both eyes), and binocular 2.5% LCLA as most important for classifying PwMS vs. controls. This composite score performed best, with AUC=0.89 (95% CI 0.85, 0.93), sensitivity=81% and specificity=80%. The composite score ROC curve performed better than any of the individual measures from the model (p<0.0001). GCIPL IED remained the best single discriminator of unilateral ON history among PwMS (AUC=0.77, 95% CI 0.71,0.83, sensitivity=68%, specificity=77%). SVM analysis performed comparably to standard logistic regression models. CONCLUSIONS:A composite score combining visual structure and function improved the capacity of SD-OCT to distinguish PwMS from controls. GCIPL IED best distinguished those with history of unilateral ON. SVM performed as well as standard statistical models for these classifications. CLASSIFICATION OF EVIDENCE/METHODS:The study provides Class III evidence that SD-OCT accurately distinguishes multiple sclerosis from normal controls as compared to clinical criteria.

On September 3^rd, 2020, the Collaborative Community on Ophthalmic Imaging (CCOI) conducted its first two-day virtual workshop on the role of artificial intelligence (AI) and related machine learning (ML) techniques for the diagnosis and treatment of various ophthalmic conditions. In a session entitled, "Artificial Intelligence For Glaucoma", a panel of glaucoma specialists, researchers, industry experts, and patients convened to share current research on the application of AI to commonly used diagnostic modalities including fundus photography, optical coherence tomography imaging, standard automated perimetry, and gonioscopy. The conference participants focused on the use of AI as a tool for disease prediction, highlighted its ability to address inequalities, and presented the limitations and challenges to its real-world clinical application. The panelists' discussion addressed AI and health equities from the clinical, societal and regulatory perspectives.

Purpose : Multiple sublayers of retina can be visualized with visible light (vis-) OCT.However, image quality can be compromised due to patient movement, cataracts, small pupil size, and light scattering causing haziness and variability in signal to noise ratio in individual A-scans and in entire B-scans.The purpose of this study was to examine the effect of conventional and deep neural network dehazing techniques on the visibility and quantitative assessment of retinal sub-layers on vis-OCT images. Methods : 9 healthy and 5 glaucoma subjects were scanned 3 times during one session.Scanning was done on the superior nasal side of para-foveal region,1.5 mm from the fovea with a 3D speckle reduction raster scanning protocol(3x3x1.6 mm with 8192x16x1024 samplings) using a prototype vis-OCT system.16 A-scan lines were averaged to reduce speckle noise.Gray-scale image dehazing guided by depth information and pretrained Dehazenet deep model following deep convolutional neural network with residual learning(DnCNN) were applied on original B-scans.Quality improvement were evaluated using quality index(QI) and contrast to noise ratio(CNR) on dehazed B-scans.For each subject, the dehazed B-scan of Dehazenet and DnCNN from a fixed location adjacent to the fovea were selected.The distances between each of 3 bright inner plexiform layers(IPL) and retinal pigment epithelium(RPE) sublayers were segmented manually for thickness measurements using a 8 A-scan averaged profile(Fig.).Coefficient of variations (CVs) were calculated to assess the measurement repeatability of the sublayers on original and dehazed B-scans. Results : Healthy and glaucoma subjects were age 45.67+/-11.7and 59.60+/-13.4(p=0.07,t-test),visual field mean deviation(MD)-1.55 to1.20 dB,and from -26.42 to -7.70dB(p= 0.003,Wilcoxon),global mean circumpapillary retinal nerve fiber layer(RNFL)thickness 96.33+/-12.20 and 59.80+/-9.09mm(p<0.001,Wilcoxon),respectively.Dehazed B-scans obtained by deep models have statistically significant better QI and CNR(Table1).Overall intra-subject CVs showed significantly improved reproducibility on all measured sub-layers of dehazed B-scans compared to original scans for all subjects(Tables 2,3). Conclusions : Vis-OCT image quality can be improved using deep neural network dehazing model resulting in higher reproducible thickness measurements of retinal sublayers within subjects in dehazed B-scans

Purpose : Deep learning of optical coherence tomography (OCT) may help discriminate glaucomatous eyes from healthy controls. However, the underlying decision making processes remain unclear. Recently, through computing class activation maps, our feature agnostic artificial intelligence of OCT images using a 3D convolutional neural network identified the optic nerve head (ONH) and its surrounding regions as structures significantly associated with glaucoma classification (PMID: 31260494). To pursue their contributions further, here we analyzed the optic nerve morphology from OCT and magnetic resonance imaging (MRI) in a subset of glaucoma and healthy subjects. Methods : Nine early glaucoma, 12 advanced glaucoma, and 4 healthy control subjects underwent spectral-domain OCT at 30x30x2 mum³ and 3-Tesla anatomical MRI at 1x1x1mm³ . Maximum intensity projection was applied to en-face OCT scans at the ONH (Fig. 1). The areas of the ONH [inner regions of interest, (ROI)] and surrounding regions (outer ROIs) visible in OCT were measured using global thresholding in ImageJ. One-way ANOVAs with post-hoc Tukey's tests were performed on the inner and outer ROIs between the 3 groups. Also, a Pearson correlation analysis was performed between the ROI areas in OCT and optic nerve volume extracted from MRI between the eye and optic chiasm. Results : For OCT of the ONH, significant group effect was observed for the areas in the inner ROIs (ANOVA: F= 7.823, p=0.00133). Post-hoc analyses revealed a significant difference between healthy controls and advanced glaucoma (p=0.0082) and between early and advanced glaucoma (p=0.0057) but no significance between healthy controls and early glaucoma (p=0.80) (Fig. 2A). No significant group effect was observed in the outer ROIs (ANOVA: F=0.004, p=0.996) (Fig. 2B). There was a negative correlation between the inner ROI area in OCT and optic nerve volume in MRI (R=-0.47, p=0.0011) (Fig. 2C). Conclusions : The ONH tissues visible on OCT appeared to contribute more than their surrounding regions to distinguishing between glaucomatous eyes and healthy eyes. The negative correlation between ONH area in OCT and optic nerve volume in MRI suggested the need to further understand the interactions between ONH and deeper brain structures in glaucoma. (Figure Presented)

Purpose : The lamina cribrosa (LC) is hypothesized to be the site of initial axonal damage in glaucoma with the peripapillary retinal nerve fiber layer (RNFL) thickness is widely used as a standard metric for quantifying this damage. The purpose of this study was to determine in vivo changes in the microstructure of the LC in eyes of non-human primates (NHP) with naturally occurring RNFL thinning. Methods : Spectral-domain OCT scans (Leica, Chicago, IL) of the optic nerve head (ONH) were acquired in vivo from a colony of 50 adult rhesus monkeys, suspected of having high prevalence of naturally occurring glaucoma. The circumpapillary global and quadrant RNFL thickness was analyzed using a custom automated segmentation software. From the set of 100 eyes, the 10 eyes with the thinnest global RNFL values were selected as the study group, while 10 eyes with RNFL values around the 50 percentile were used as the control group. A previously described automated segmentation algorithm was used for LC microstructure analysis. The LC microstructure was analyzed globally and in the th following volumetric sectors: quadrants, central and peripheral lamina, and 3 depth slabs (anterior, middle, posterior; Figure). Beam thickness/pore diameter ratio (BPR) and connective tissue volume fraction (CTVF: beam volume/total volume) were calculated globally and in sectors. Results : 20 eyes (15 animals) were analyzed (Table 1). While no significant difference was detected between groups for age, weight or disc size, the study group had significantly thinner RNFL than the control group (p<0.01). The study group had significantly larger BPR and CTVF compared with the control group (Table 2). Significant sectoral differences between study and control group RNFL thickness were noted for BPR and CTVF in the nasal and temporal quadrants, central LC, and in LC depth. Across eyes, the global RNFL thickness was moderately negatively correlated only with the global CTVF (lower RNFL thickness associated with higher CTVF; r² =0.63, p=0.045). Conclusions : Eyes with thinner circumpapillary RNFL had thicker LC BPR and CTVF globally and in various sectors when compared to eyes with normal RNFL thickness. Whether these LC changes are the cause of RNFL damage or the result of remodeling of the LC requires further investigation. (Figure Presented)

Purpose : Automated segmentation of in-vivo lamina cribrosa (LC) has been challenging, owing to the complex 3D structure and decreased visibility in the lamina depth. Frangi's vesselness filter, which was originally developed for angiogram segmentation, have been successfully demonstrated in segmenting the ex-vivo LC from micro-CT and second harmonic generation microscopy images. In this project we are proposing a new approach of segmenting the in vivo LC from OCT scans, incorporating the Frangi's vesselness principle to facilitate in vivo LC image analysis in much greater detail compared to our previously described 3D analysis method. Methods : In-vivo spectral-domain OCT scans (Leica, Chicago, IL) were acquired from healthy non-human primates. Scans of varying degree of image quality were selected for the analysis and underwent automated brightness and local contrast enhancement. 3D Frangi's vesselness filter was applied using a fixed setting for scans of all qualities. Our previously described segmentation algorithm was then used to quantify the LC microstructure. The measurements generated from the Frangi analysis and from our own conventional method were compared with a standard reference (manually segmented LC by an expert). Paired t tests were performed to compare if the differences between standard reference and conventional method are greater than the differences between standard reference and Frangi analysis. The visibility of analyzable lamina and dice coefficient were also compared to the conventional method using the same test. Results : In vivo scans acquired from 5 rhesus macaques (3 males, 1 female, aged 4.3-10.7 yrs) were used for the analysis. No significant difference was detected for LC microstructure parameters between Frangi's approach and conventional method with respect to the standard reference, except for significantly higher pore count in Frangi's method (p=0.003; Table). Furthermore, visibility (Figure) was significantly higher for the Frangi method compared to the conventional approach (p<0.001) with no difference detected for the semantic segmentation, as reflected by the dice coefficient. Conclusions : The use of Frangi analysis substantially increase the analyzable lamina while providing similar quantification of the LC microstructure compared to our previous 3D analysis method. This improves the potential for automated and thorough volumetric analysis of in vivo OCT LC image

Purpose : Glaucoma is an eye disease with widespread involvement of the brain. Since visual cortex (VC) may possess lower choline levels in glaucoma, and basal forebrain (BF) has cholinergic projections to VC for modulating cerebral blood flow and visual processing, we postulate that the vascular functions of the VC and BF are involved in glaucoma (PMID: 31242454). Recently, we used a novel whole-brain relative cerebrovascular reactivity (rCVR) mapping technique via resting-state functional MRI (rsfMRI) without gas challenge, and observed rCVR decrease in VC and rCVR increase in BF in patients with increasing glaucoma severity (PMID: 34892116). However, the underlying mechanisms remain to be elucidated. Here, we applied a hydrogel-induced glaucoma mouse model to elevate intraocular pressure (IOP) (PMID: 31176841), mapped wholebrain rCVR using rsfMRI, and measured optomotor responses (OMR). We hypothesize that chronic IOP elevation can lead to rCVR changes in the glaucomatous brain along with visual impairments. Methods : For the glaucoma model, C57BL/6J mice (male, 15-weeks, n=15) received intracameral injection of cross-linking hydrogel to the right eye to obstruct aqueous outflow and induce chronic IOP elevation. Controls (male, 15-weeks, n=13) were untreated. IOP was measured in both eyes 2-3 times per week for 3 weeks, followed by OMR and rsfMRI experiments at 7 Tesla (Fig. 1A). Results : Sustained IOP elevation was confirmed in the right eyes of the glaucoma model (Fig. 1B). Over 90% of mouse optic nerve fibers are known to project to the contralateral visual brain; rCVR decreased in the left but not right VC, whereas rCVR increased in the right BF in the glaucoma model but not the controls (Fig. 2A). These rCVR changes were inversely coupled (Fig. 2B). In addition, IOP of the injected eye was inversely correlated with rCVR in the left VC, while positively correlated with rCVR in the right BF (Fig. 2C). OMR revealed a decrease in visual acuity and an increase in visual contrast threshold for the injected eye (Fig. 2D) indicating visual impairment. The decrease in visual acuity was inversely correlated with rCVR in the BF (Fig. 2E). Conclusions : Mouse rCVR mapping using rsfMRI detects widespread brain changes induced by chronic IOP elevation, and demonstrates vascular involvement in glaucoma both within and beyond the primary visual pathways

Purpose : The ability to detect progression in eyes with advanced glaucoma is challenging because of known limitations of commonly used structural and functional parameters reaching their minimal measurable limit (floor effect) or increased measurement variability. We examined the ability of inner nuclear layer (INL) thickness measurements to demonstrate change longitudinally in eyes with early and advanced severity glaucoma. Methods : Subjects with glaucoma and >=4 visits were included in the study. Subjects in the ?Early/Moderate? group (EG) had average circumpapillary retinal nerve fiber layer (cRNFL) thicknesses >=60mum and subjects in the ?Advanced? group (AG) had average cRNFL thicknesses <=60mum. All subjects had comprehensive ophthalmic examination, Humphrey visual field (Zeiss, Dublin, CA) testing, and spectral-domain OCT (Cirrus HD-OCT; Zeiss) optic nerve head (ONH) and macula scans. Segmentation of the INL was performed using the Iowa Reference Algorithms (Retinal Image Analysis Lab, Iowa Institute for Biomedical Imaging, Iowa City, IA) and segmentation errors were manually corrected by a trained grader. Overall INL thickness along with the superior and inferior hemifields were used for analysis. Rates of progression were estimated from longitudinal OCT and visual field (VF) data using mixed effects models adjusting for baseline age, follow-up duration, and signal strength at each visit. Results : 23 eyes (23 subjects), 12 with EG and 11 with AG, were included in the study. At baseline, a statistically significant difference between groups was detected in MD, cRNFL, and GCIPL thicknesses (Table 1). In EG eyes, the rate of change was significantly different than a zero slope for cRNFL thickness, C:D ratio, and GCIPL thickness (Table 2). Inferior INL thickness was the only INL parameter showing significant rate of change. However, in the advanced group, all parameters (including both global and sectoral INL thicknesses) showed significant rate of change except for the cRNFL. Conclusions : Longitudinal measurements of INL thickness may be useful for following disease progression in subjects with advanced-stage glaucoma where cRNFL thickness is no longer useful

Purpose : Glaucoma is a widespread neurodegenerative disease affecting the retinal ganglion cells, optic nerve, distal visual pathways and beyond. Recent studies suggest that cerebrospinal fluid (CSF) plays a role in clearing wastes from the brain and that CSF dynamics may be altered in neurodegenerative diseases. Since CSF dynamics can be facilitated by the global brain activity, in the present study, we investigated how the dynamics of CSF and its coupling with global brain activity may be altered in glaucoma using functional magnetic resonance imaging (fMRI). Methods : 19 early glaucoma patients (62.3+/-1.7 yrs) (mean+/-SEM), 19 advanced glaucoma patients (64.7+/-2.4 yrs), and 19 healthy subjects (59+/-2.4 yrs) underwent anatomical MRI and resting-state fMRI with eyes closed. Age did not differ across groups (P=0.188). We extracted the CSF signal time profiles from the fourth ventricle (Fig. 1A) and the global brain activity [blood-oxygenation-level-dependent signal time profiles] from the entire gray matter (Fig. 1B). Following previous literature (Han F, et al. PLOS Biol 2021;19), the coupling between the CSF signals and the global brain activity (CSF-BOLD coupling) was examined via cross correlation at the 4s time lag, where more negative values indicate stronger coupling. We also associated these correlations with the volumes of the anterior visual pathway in anatomical MRI. Results : A significant group difference was observed in the power (i.e., strength) of the low frequency (0.01-0.03Hz) in the CSF signals (P=0.013; Fig.1C). Specifically, early glaucoma patients showed significantly greater power than advanced glaucoma patients (Bonferroni P=0.010). The power of the global brain activity showed similar trends but did not reach significance (P=0.390; Fig.1D). The CSF-BOLD coupling at the 4s lag differed significantly across groups (P=0.007; Fig. 1E). Early glaucoma patients had significantly stronger coupling than advanced glaucoma patients (Bonferroni P=0.025) and healthy controls (Bonferroni P=0.013). Further, CSF-BOLD coupling was correlated with the volumes of optic nerve (right: R=-0.342, P=0.009; left: R=-0.344, P=0.009, Fig. 2D,E) and optic chiasm (R=0.264, P=0.047, Fig. 2F). Conclusions : Our observations of the altered CSF dynamics and CSF-BOLD coupling provide physiological evidence to support the recent hypothesis of widespread brain involvements in the early stage of glaucoma

Purpose : Previously we described the longitudinal glaucoma relationship between structure and function using a broken stick analysis approach to identify the location where the rate of change accelerates or decelerates. In that analysis we used each measurement point as an independent point, aggregated all eyes from all visits, and treated longitudinal data as cross-sectional. Using improved statistical methodology, we accounted for repeated measurements and the use of data from both eyes in the longitudinal model. The purpose of this study is to identify the locations of tipping points and rates of change before and after them in structural and functional measurements. Methods : Subjects with comprehensive ophthalmic examination and 5 or more visits with qualified visual fields (VF; Humphrey Field Analyzer; Zeiss, Dublin, CA) and OCT (Cirrus HD-OCT; Zeiss) with ONH and macular scans were enrolled. Segmented mixed models that account for repeated measurements were utilized to estimate the tipping points and the difference-in-slope. The number of tipping points was determined by identifying the optimal model using Bayesian information criterion. Results : 216 eyes (164 open angle glaucoma, 45 glaucoma suspect, and 7 healthy eyes) of 145 subjects were analyzed (Table). Retinal nerve fiber layer (RNFL), and ganglion cell inner retinal layer (GCIPL) decreases and cup to disc ratio (CDR) increases since early stages of the disease were measured (Figure). Unlike previous cross-sectional reports, visual field mean deviation (MD) also decreases along with structural parameters since early stages of the disease. RNFL thinning stalls beyond MD<-15.63dB (Figure A) while GCIPL keeps decreasing (B), and CDR slowly increases (C) throughout the functional damage range. Direct comparison between the structural parameters shows that RNFL thinning decelerates in advanced disease compared to both GCIPL and CDR and GCIPL thinning decelerates compared to CDR. Conclusions : Structural and functional measurements (RNFL, GCIPL, CDR and MD) are useful to evaluate glaucoma change from early stages of the disease. As glaucoma progresses and RNFL reaches its minimal measurable level GCIPL, CDR and MD remain useful to evaluate the disease. The clinical routine for following subjects with glaucoma should account for the ability to measure relevant parameters at various stages of disease