Faculty Bibliography

Purpose : The uncertainty quantification of segmentation results is critical for understanding the reliability of the segmentation model. The purpose of this study is to investigate the effect of deep learning-based segmentation with uncertainty measurement in the relationship between RNFL thickness and visual field mean deviation (MD) Methods : Optical coherence tomography (OCT) scans were acquired from both eyes on 634 glaucoma patients, 404 glaucoma suspects, and 49 healthy controls using commercial OCT device (Cirrus HD-OCT, 200x200 Optic Disc Cubes; Zeiss, Dublin, CA). All subjects had visual field (VF) tests at each visit (Humphrey VF, SITA 24-2 test; Zeiss). A segmentation model was trained using Bayesian deep learning for voxel-wise segmentation of RNFL layer in OCT volume and compute the voxel-wise uncertainty of the segmentation output. The higher uncertainty denotes the unreliability of the segmentation and vice versa and it allows the determination of erroneous segmentation at test time. Uncertainty-guided global mean of the RNLF thickness (RNFL-Umean) was then computed by discarding the voxels with erroneous segmentation labels with higher uncertainty during the thickness computation. Also, the global mean of the RNLF thickness (RNFLmean;) was computed without taking uncertainty into account. Pearson correlation coefficient between RNFLU and MD was computed and compared with the Pearson correlation coefficient between RNFLmean;and MD. Results : The proposed RNFL-Umean;gave stronger correlation with MD than RNFLmean;The Pearson correlation coefficients were (0.67 (RNFL-Umean;) vs 0.63 (RNFL mean;; p<0.001) for glaucoma subjects, (0.56 vs 0.53 ;p=0.01) for glaucoma suspects and (0.08 vs 0.01; p=0.21) for normal subjects. Conclusions : The proposed uncertainty-guided computation of RNFL thickness showed improved correlation with the visual field MD. This demonstrates that segmentation uncertainty can be used to reduce the effect of inaccurate segmentation in computing the RNFL thickness. This also shows that uncertainty-guided computation of RNFL thickness is a better predictor of visual function than the normal RNFL thickness computed without using uncertainty

Purpose : The lamina cribrosa (LC) is considered to play an important role in the pathogenesis of glaucoma. In this study, we investigate the shape variation (SV) of the LC pores as a potential biomarker for quantifying the morphological irregularity in vivo. Methods : 36 healthy and 14 glaucomatous (GL) eyes (total: 39 subjects) underwent a comprehensive ophthalmic examination and scanning of the optic nerve head with sweptsource OCT (Table 1). Images were converted to isotropic and pores were segmented using ImageJ. SV was defined as the mean-squared error of the pore pattern with respect to a solid circle (Figure 1(a)) with SV of a circle marked as zero, and higher SV value with increasing shape irregularity. SV of each pore was automatically calculated by a MATLAB code. The overall SV value was generated as the average of SV in the stack of individual slices. Age effect on SV was examined in all healthy eyes and a subset of 14 eyes was selected for age-matched comparison with the glaucomatous eyes (Table 1). Results : No significant correlation was detected between SV and age (p=0.145; Spearman correlation) in all healthy subjects. Examining the effect of depth on the difference between SV of glaucomatous and healthy eyes, the posterior third of the LC had significantly lower than other sections (p=0.007; Figure 1(b)). SV in glaucoma eyes was significantly higher than in the healthy group (p=0.008; Figure 1(c)). Conclusions : We demonstrated morphological differences in pore shape variation between healthy and glaucoma eyes that is mostly affecting the anterior 2/3 of the LC. Further studies are warranted to assess the use of SV as a structural biomarker in glaucoma

Purpose : Mutations in optineurin (OPTN) are associated with familial normal tension glaucoma and other neurodegenerative diseases. It remains unclear how OPTN loss or mutation alters visual function during aging. Here, we used transgenic mouse models and in vivo assessments to test the hypothesis that OPTN dysfunction contributes to progressive visual impairment through a toxic gain of function mechanism. Methods : Mice with C57BL/6 background were used (Fig 1): wildtype (WT; n=19), homozygous OPTN knock-out (mOPTN-KO; n=13), hemizygous mouse E50K OPTN knock-in (mE50K-het; n=8), homozygous mouse E50K OPTN knock-in (mE50K homoz; n=10), and human E50K OPTN bacterial artificial chromosome overexpression (hE50K BAC; n=6) (PMID: 31076632, 25818176). Intraocular pressure (IOP), total retinal thickness (TRT), visual acuity (VA), and contrast sensitivity (CS) were measured at 6, 12, and 18 months of age in the same mice using the TonoLab rebound tonometer, Bioptigen spectral-domain optical te uses cookies. By continuing to use our website, you are agreeing to coherence tomography imaging, and OptoMotry optokinetic virtual reality system respectively. Left and right eye data were averaged and analyzed using ANOVAs followed by posthoc tests between genotype and age groups, as well as linear regressions for VA versus contrast threshold (CT). Results : Our longitudinal study of the same mice during the aging process showed that IOP remained normal between 10-15 mmHg (Fig 2A). Small to no difference in TRT over time or compared to WT was observed (Fig 2B). mE50K-homoz, mE50K-het, and hE50K BAC mice exhibited greater age-dependent decline in VA and CT than WT or mOPTN-KO mice (Fig 2C, 2D, 2E). In contrast, mOPTN-KO mice showed preservation of VA and CT over time compared to WT. Consistently, mice with one copy of E50K OPTN (mE50K het) experienced less deterioration of VA and CT compared to mice with two copies (mE50K homoz) or mild overexpression (hE50K BAC). Conclusions : Depsite limited IOP and TRT changes between age and genotype groups, E50K OPTN was associated with differential age-dependent visual impairment (greater for CS than VA). Surprisingly, OPTN deficiency preserved visual function such that CS in knockout mice was better than WT mice. Our results suggest visual loss associated with E50K OPTN is due to a toxic gain of function mechanism, and that suppression of OPTN might constitute a therapeutic strategy for glaucomatous neurodegeneration

Purpose : Rhesus macaques are a common animal model in ophthalmology because of the high similarity of their eyes and visual pathway to human. The characterization of optic nerve head (ONH) and peripapillary region in monkeys reported so far mostly involved a manual process which is laborious and subjected to operator errors. It is also usually generated from a cohort of similar age group. In this cross-sectional observational study, we deploy automated and manual segmentations to evaluate the OCT retinal nerve fiber layer (RNFL) thickness, ONH and lamina cribrosa (LC) microstructure parameters in a cohort of free roaming macaques. Methods : In-vivo ONH spectral-domain OCT scans (Leica, Chicago, IL) were obtained by a single experienced operator after excluding eyes with any retinal pathologies. The margins of the optic disc were drawn manually and the resultant scans were analyzed using an automated segmentation software of our own design. The LC microstructure parameters were obtained through a previously described segmentation algorithm. The other parameters of ONH, namely the cup-to-disc (C/D) ratio and minimum rim width (MRW) were assessed manually. Wilcoxon rank sum test was used to test the association of LC parameters, C/D ratio and MRW with age, while the rest of the parameters were analyzed using mixed effects model accounting for age, sex and intra-subject correlation. Results : 29 eyes from 19 monkeys (11 females, 8 males) with age ranging from 4.2 to 23.8 years were analyzed. Males were overall bigger and significantly heavier than females in our cohort (Table 1). Superior RNFL was thicker in male and is the only RNFL parameter that was associated with age or sex in this healthy cohort. No significant association was detected for any of the ONH parameters with age or sex. LC was more visible and thicker in male with higher beam to pore ratio and connective tissue fraction than in female. Conclusions : The characterization of normal macaque eyes from a cohort of free roaming animals is useful as a standard reference to assess pathological changes in future experimental studies

Purpose : Schlemm canal (SC) is characterized by high local variations in morphology. Previously, we reported characteristics of SC using SC area measurements by optical coherence tomography (OCT) in healthy eyes. Herein, we examine the interobserver variability of SC height, width, and area in glaucomatous and healthy eyes. Methods : The anterior segment of six eyes from three subjects (1 female, 2 male) were imaged using OCT (Cirrus HD-OCT, Zeiss, Dublin, California, USA). A 4x4mm volumetric image of the limbus (depth of 2mm) was acquired with the Anterior Segment Cube scan protocol, comprised of 128 horizontal B-scans composed of 512 A-scans. SC was positioned to the side of the image to maximize visualization of aqueous humor vessel crossings. Scans were processed to maximize visualization of SC; image volumes were averaged (3x3x3 kernel) and contrast was enhanced with the local histogram algorithm using Fiji (version 2.10/1.53c). A cross-sectional B-scan and the two B-scans +/-5 frames were identified as three reference frames, based on best visualized SC location (Fig. 1). Three independent observers performed manual segmentation to measure SC width, height, and cross-sectional area for these three reference frames per volume. Width was defined as the longest measure of SC and height as perpendicular to the line used for width measurement. The observers performed these measurements on 15 volumes for a total of 45 analyzed frames each. The coefficient of variation was calculated based on standard deviations estimated using hierarchical multi-level random-effects models. Interobserver variability was quantified with a two-way ANOVA to calculate the intraclass correlation coefficient (ICC). Results : Participants had a mean age of 72.0 +/- 7.47 years (range: 66 to 82) and consisted of one healthy subject and two with primary open angle glaucoma. Measurement means and variation are presented in Table 1. The ICCs for interobserver variability are excellent for width measurements and low to moderate for height and area (Table 2) Conclusions : Excellent ICC for interobserver variability of SC width suggests it is suitable for use in clinical trials

Purpose : To improve the performance of the feature agnostic AI-based glaucoma detection algorithm by evaluating an uncertainty score for each prediction. Methods : We previously developed a 5-layer 3D Convolutional Neural Network (CNN) in using the OCT scans from both eyes of 134 healthy, 779 glaucoma patients on a Cirrus HDOCT scanner (200x200 ONH Cubes; Zeiss, Dublin CA). In our analysis, we excluded scans with signal strength less than 7 and downsampled the volumes to 64x64x128 voxels. Uncertainty of AI models can be estimated by computing the effect of randomly ignoring a set of parameters within the network. We randomly zeroed 5% of each of the 5 convolutional layers and computed the entropy in the final score over 20 forward passes. The performance of the approach was assessed using a 10-fold cross validation study. Results : Over the 10-folds, the model showed an AUC of 0.91+/-0.027. In analysing the uncertainty and the probabilistic scores generated by the model (Softmax function) for one fold (see Fig. 1), we observed that a threshold of 0.8 can be used to flag 75% of the false positives and false negatives for further review. On the other hand, only 25% of the healthy controls and 20% of glaucoma patients showed an uncertainty score above that threshold. Fig. 2 summarises the overall uncertainties scores and indicates that low scores are associated with the correctly identified cases while the errors show higher uncertainty scores. Conclusions : The quantitative uncertainty measure provides supplementary information to clinicians and can be used to flag difficult cases automatically. Given that the dataset used in this work is highly imbalanced (more positive cases compared to normal cases) the uncertainty score for true negative cases is significantly higher compared to true positive cases. We expect to achieve lower uncertainty scores for normal cases if more data for normal eyes are available. The uncertainty analysis presented here may aid clinical interpretations of AI-based glaucoma detection outcomes. A separate study will be run to measure this improvement and compare the result with experts' level of uncertainty

OBJECTIVE:Familial Dysautonomia (FD) disease, lacks a useful biomarker for clinical monitoring. In this longitudinal study we characterized the structural changes in the macula, peripapillary and the optic nerve head (ONH) regions in subjects with FD. METHODS:Data was consecutively collected from subjects attending the FD clinic between 2012 and 2019. All subjects were imaged with spectral-domain Optical Coherence Tomography (OCT). Global and sectoral measurements of mean retinal nerve fiber layer (RNFL) and macular ganglion cell and inner plexiform layer (GCIPL) thickness, and ONH parameters of rim area, average cup-to-disc (C:D) ratio, and cup volume were used for the analysis. The best fit models (linear, quadratic and broken stick linear model) were used to describe the longitudinal change in each of the parameters. RESULTS:91 subjects (149 eyes) with FD of ages 5-56 years were included in the analysis. The rate of change for average RNFL and average GCIPL thicknesses were significant before reaching a plateau at the age of 26.2 for RNFL and 24.8 for GCIPL (- 0.861 µm/year (95% CI - 1.026, - 0.693) and - 0.553 µm/year (95% CI - 0.645, - 0.461), respectively). Significant linear rate of progression was noted for all ONH parameters, except for a subset of subjects (24%), with no cupping that did not show progression in any of the ONH parameters. CONCLUSIONS:The rapidly declining RNFL and GCIPL can explain the progressive visual impairment previously reported in these subjects. Among all structural parameters, ONH parameters might be most suitable for longitudinal follow-up, in eyes with a measurable cup.

Purpose/UNASSIGNED:The purpose was to establish the position of the fovea centralis to the optic nerve via en-face, near-infrared spectral domain optical coherence tomography (NIR-OCT) in healthy patients. This may shed light on physiological variability and be used for studying subtle cases of foveal ectopia in macular pathology and after retinal detachment. Methods/UNASSIGNED:SD-OCT data of 890 healthy eyes were retrospectively analyzed. Exclusion criteria included axial myopia causing tilting of the optic disc, peripapillary atrophy >1/3 the width of the disc, macular images excluding greater than half of the optic disc, and patients unable to maintain vertical head positioning. Two independent reviewers measured the horizontal and vertical distance from the fovea to the optic disc center and optic disc diameter via cross-sectional and en-face scanning laser ophthalmoloscopy OCT imaging. Results/UNASSIGNED:890 eyes were included in the study. The right and left eyes differed in the horizontal distance from the fovea to the disc center (4359 vs. 4248 µm, P < 0.001) and vertical distance from the fovea to the disc center (464 µm vs. 647, P < 0.001). This corresponded to a smaller angle between the right and left eyes (6.07° vs. 8.67°, P < 0.001). Older age was associated with a larger horizontal (P = 0.008) and vertical distance (0.025). These differences persisted after correcting for axial length in the 487 patients with axial-length data. Conclusions/UNASSIGNED:This study compares the position of the fovea centralis among individuals without macular pathology on a micron level basis. The significant variability between right and left eyes indicates that contralateral eye evaluation cannot be reliably used. Rather, true foveal ectopia requires assessments of preoperative and postoperative NIR-OCT scans. This finding is relevant to retinal detachment cases and evaluation of subtle foveal ectopia. Translational Relevance/UNASSIGNED:This finding is relevant to retinal detachment cases and evaluation of subtle foveal ectopia.