Structure-function relationships in glaucoma using enhanced depth imaging optical coherence tomography-derived parameters: a cross-sectional observational study

Peripheral Microvascular Abnormalities Associated with Open-Angle Glaucoma

PURPOSE

To investigate peripheral microvascular abnormalities associated with patients with open-angle glaucoma (OAG).

DESIGN

This was a cross-sectional study.

PARTICIPANTS

Patients with OAG and controls.

METHODS

All subjects underwent detailed ophthalmic evaluation, including Humphrey visual field (HVF) tests and swept source OCT. To evaluate peripheral microvascular abnormalities, nailfold capillaroscopy (NFC) and laser Doppler imaging (LDI) were performed. The presence of microhemorrhages, tortuous capillaries, dilated capillaries, avascular areas, and the capillary density, among other characteristics, were recorded using NFC; fingertip blood flow (FBF) was measured using LDI at different time points, before and 1, 10, and 20 minutes after exposure to a cold stimulus. In addition, venous blood samples were collected to measure serum endothelin-1 (ET-1) concentrations as well as serum autoantibodies.

MAIN OUTCOME MEASURES

Presence of microhemorrhages, tortuous capillaries, and dilated capillaries; FBF; ET-1; and autoantibodies.

RESULTS

Sixty-eight subjects (43 patients with OAG and 25 controls) were enrolled in the study. Microhemorrhages were found in the nail bed of 65.1% of the patients with OAG compared with 25.0% of the controls (P = 0.003). There was a significant difference in the mean FBF at the baseline in patients with OAG versus controls (293.6 ± 100.2 vs 388.8 ± 52.0 perfusion units, respectively, P < 0.001), together with a significant decrease in the mean FBF 10 and 20 minutes after cold stimulus in patients with OAG in comparison to controls (P < 0.001 for all comparisons). There was a positive correlation between mean baseline FBF and HVF mean deviation (r = 0.27, P = 0.03) and between mean baseline FBF and average retinal nerve fiber layer thickness (r = 0.44, P = 0.001). Neither the analysis of ET-1 concentrations (P= 0.71) nor the autoantibodies measurements (P > 0.05, for all) showed any difference between the 2 groups.

CONCLUSIONS

Significant peripheral microvascular abnormalities were found in patients with OAG compared to controls, suggesting that microvascular changes might play a role in the pathogenesis of the disease. In addition, part of these peripheral microvascular abnormalities seems to be correlated with both functional and structural glaucomatous damage.

FINANCIAL DISCLOSURE(S)

The author(s) have no proprietary or commercial interest in any materials discussed in this article.

Detecting glaucoma from multi-modal data using probabilistic deep learning

Objective

To assess the accuracy of probabilistic deep learning models to discriminate normal eyes and eyes with glaucoma from fundus photographs and visual fields.

Design

Algorithm development for discriminating normal and glaucoma eyes using data from multicenter, cross-sectional, case-control study.

Subjects and participants

Fundus photograph and visual field data from 1,655 eyes of 929 normal and glaucoma subjects to develop and test deep learning models and an independent group of 196 eyes of 98 normal and glaucoma patients to validate deep learning models.

Main outcome measures

Accuracy and area under the receiver-operating characteristic curve (AUC).

Methods

Fundus photographs and OCT images were carefully examined by clinicians to identify glaucomatous optic neuropathy (GON). When GON was detected by the reader, the finding was further evaluated by another clinician. Three probabilistic deep convolutional neural network (CNN) models were developed using 1,655 fundus photographs, 1,655 visual fields, and 1,655 pairs of fundus photographs and visual fields collected from Compass instruments. Deep learning models were trained and tested using 80% of fundus photographs and visual fields for training set and 20% of the data for testing set. Models were further validated using an independent validation dataset. The performance of the probabilistic deep learning model was compared with that of the corresponding deterministic CNN model.

Results

The AUC of the deep learning model in detecting glaucoma from fundus photographs, visual fields, and combined modalities using development dataset were 0.90 (95% confidence interval: 0.89-0.92), 0.89 (0.88-0.91), and 0.94 (0.92-0.96), respectively. The AUC of the deep learning model in detecting glaucoma from fundus photographs, visual fields, and both modalities using the independent validation dataset were 0.94 (0.92-0.95), 0.98 (0.98-0.99), and 0.98 (0.98-0.99), respectively. The AUC of the deep learning model in detecting glaucoma from fundus photographs, visual fields, and both modalities using an early glaucoma subset were 0.90 (0.88,0.91), 0.74 (0.73,0.75), 0.91 (0.89,0.93), respectively. Eyes that were misclassified had significantly higher uncertainty in likelihood of diagnosis compared to eyes that were classified correctly. The uncertainty level of the correctly classified eyes is much lower in the combined model compared to the model based on visual fields only. The AUCs of the deterministic CNN model using fundus images, visual field, and combined modalities based on the development dataset were 0.87 (0.85,0.90), 0.88 (0.84,0.91), and 0.91 (0.89,0.94), and the AUCs based on the independent validation dataset were 0.91 (0.89,0.93), 0.97 (0.95,0.99), and 0.97 (0.96,0.99), respectively, while the AUCs based on an early glaucoma subset were 0.88 (0.86,0.91), 0.75 (0.73,0.77), and 0.92 (0.89,0.95), respectively.

Conclusion and relevance

Probabilistic deep learning models can detect glaucoma from multi-modal data with high accuracy. Our findings suggest that models based on combined visual field and fundus photograph modalities detects glaucoma with higher accuracy. While probabilistic and deterministic CNN models provided similar performance, probabilistic models generate certainty level of the outcome thus providing another level of confidence in decision making.

Diagnostic Performance of Optical Coherence Tomography for Pseudoexfoliation Glaucoma

PRECIS

LC thickness and LCCI had comparable diagnostic performances with RNFL thickness in distinguishing eyes with PXG from those with PXS. BMO-MRW showed the lowest diagnostic performance among all geometric parameters derived from OCT scans we evaluated.

OBJECTIVE

To compare the diagnostic performance of different geometric parameters derived from optical coherence tomography (OCT) scans (retinal nerve fiber layer [RNFL] thickness, lamina cribrosa [LC] thickness, LC curvature index [LCCI] and Bruch's membrane opening-minimum rim width [BMO-MRW]) for distinguishing eyes with pseudoexfoliation glaucoma (PXG) from pseudoexfoliation syndrome (PXS) and healthy eyes.

METHODS

Fifty-five eyes of 55 patients with PXG, 55 eyes of 55 patients with PXS, and 50 healthy subjects were enrolled in this cross-sectional study. The areas under the receiver operating characteristic curves (AUCs) of RNFL thickness, LC thickness, LCCI and BMO-MRW were calculated and compared.

RESULTS

In discriminating between eyes with PXG from those with PXS, LC thickness (0.930 [95% CI: 0.883-0.978]) and global RNFL thickness (0.974 [95% CI: 0.947-0.992]) presented comparable AUCs (P=0.244). In distinguishing subjects wiht PXG from healthy controls, both LC thickness (0.972 [95% CI: 0.948-0.997]) and LCCI (0.983 [95% CI: 0.968-0.998]) had comparable AUCs with global RNFL thickness (0.988 [95% CI: 0.974-1.000]) (P=0.094 andP=0.239, respectively). Global BMO-MRW had lower AUCs than RNFL thickness (0.839 [95% CI: 0.759-0.920] and 0.897 [95% CI: 0.836-0.958], respectively) in distinguishing PXG from both PXS and healthy controls (P=0.001 andP=0.002, respectively). BMO-MRW also had significantly lower AUCs than both LC thickness and LCCI in distinguishing PXG from healthy controls (P=0.034 andP=0.001, respectively).

CONCLUSION

LC thickness and LCCI had better diagnostic performance than BMO-MRW in distinguishing PXG from PXS and healthy controls, which were comparable to RNFL thickness.

New views on three-dimensional imaging technologies for glaucoma: an overview

PURPOSE OF REVIEW

To summarize the literature on three-dimensional (3D) technological advances in ophthalmology, the quantitative methods associated with this, and their improved ability to help detect glaucoma disease progression.

RECENT FINDINGS

Improvements in measuring glaucomatous structural changes are the result of dual innovations in optical coherence tomography (OCT) imaging technology and in associated quantitative software.

SUMMARY

Compared with two-dimensional (2D) OCT parameters, newer 3D parameters provide more data and fewer artifacts.

Correlation Between Changes in Lamina Cribrosa Structure and Visual Field in Primary Open-Angle Glaucoma

Purpose

To investigate the correlation between changes in lamina cribrosa (LC) structure using enhanced depth imaging (EDI) and severity of visual field (VF) affection in primary open-angle glaucoma (POAG).

Patients and Methods

This prospective cross-sectional study was carried out on 52 glaucomatous eyes of 28 POAG patients who attended Tanta University Eye Hospital and 40 normal eyes of 20 age-matched normal subjects from April 2020 to March 2021. POAG patients were classified according to the modified Hodapp-Anderson-Parrish grading scale based on the MD of the standard automated perimetry (SAP) visual field into two groups: group (1) - mild-to-moderate POAG patients (MD ≤ 12 dB), group (2) - severe POAG patients (MD ≥ 12 dB) and the third group included normal (control) age- and gender-matched individuals.

Results

There was no statistically significant difference between the three groups regarding lamina cribrosa thickness nor lamina cribrosa area and there was no statistically significant correlation between MD and LC thickness, LC area (P-value=0.395 and 0.644). There was a statistically significant correlation between MD and anterior lamina cribrosa surface depth (P-value=0.002) and there was a statistically significant positive correlation between MD and prelaminar neural tissue (PLNT) thickness and prelaminar neural tissue (PLNT) area (P-value= 0.023 and <0.001, respectively).

Conclusion

EDI-OCT is a useful biomarker of structural changes in ONH and LC, and we recommended it to be a part of the routine monitoring of patients with POAG.

Peripapillary Hyper-reflective Ovoid Mass-like Structure (PHOMS): An Optical Coherence Tomography Marker of Axoplasmic Stasis in the Optic Nerve Head

BACKGROUND

With the development and widespread adoption of spectral-domain optical coherence tomography (OCT), peripapillary hyper-reflective ovoid mass-like structures (PHOMS) have become a frequent OCT finding in neuro-ophthalmic practice. Although originally assumed to represent a form of buried optic disc drusen (ODD), PHOMS differ from ODD in many important ways. The histopathological underpinnings of PHOMS are now becoming more clearly understood.

EVIDENCE ACQUISITION

Review of literature.

RESULTS

PHOMS can be broadly classified as disk edema-associated PHOMS, ODD-associated PHOMS, or anomalous disk-associated PHOMS. PHOMS are seen in many conditions, including papilledema, nonarteritic anterior ischemic optic neuropathy, central retinal vein occlusion, acute demyelinating optic neuritis, ODD, and tilted disks (myopic obliquely inserted disks) and in many cases resolve along with the underlying condition. The histopathological study of these diverse entities reveals the common feature of a bulge of optic nerve fibers herniating centrifugally over Bruch membrane opening into the peripapillary space, correlating exactly with the location, shape, and space-occupying nature of PHOMS on OCT. Because of the radial symmetry of these herniating optic nerve fibers, PHOMS are best thought of as a complete or partial torus (i.e., donut) in 3 dimensions.

CONCLUSIONS

PHOMS are a common but nonspecific OCT marker of axoplasmic stasis in the optic nerve head. They are not themselves ODD or ODD precursors, although they can be seen in association with ODD and a wide spectrum of other conditions. They do not exclude papilledema and often accompany it. The circumferential extent and characteristic 3D toroidal nature of a PHOMS are best appreciated by scrolling through consecutive OCT images.

Structural abnormalities associated with glaucoma using swept-source optical coherence tomography in patients with systemic sclerosis

PURPOSE

Vasospasm represents an early event in systemic sclerosis (SSc). Ocular vasospasm may induce optic nerve head (ONH) damage and has been involved in the pathogenesis of glaucoma, especially normal-tension glaucoma (NTG). We aimed to investigate the presence of structural abnormalities associated with NTG using swept-source optical coherence tomography (SS-OCT) and to correlate the OCT parameters with clinical, capillaroscopy and digital blood flow measures in patients with SSc.

METHODS

In this cross-sectional study, 40 patients with SSc and 23 age-matched controls were included. The following parameters were measured using SS-OCT: mean and sectoral retinal nerve fibre layer (RNFL) thickness, macular ganglion cell layer complex (GCC) thickness and ONH morphology. Nailfold capillaroscopy (NFC) and digital blood flow measurements using laser Doppler imaging (LDI) were performed in all subjects.

RESULTS

Patients with SSc showed a thinner temporal RNFL than the controls (69.23 ± 11.74 versus 83.35 ± 20.19 µm, p = 0.001). The other parameters were similar between the two groups. In SSc patients, there was an inverse correlation between the disease duration and the average, superior and inferior RNFL thickness and the GCC thickness and between Raynaud's phenomenon duration and the average RNFL and GCC thickness (p < 0.05). NFC and LDI measurements did not show correlations with OCT parameters.

CONCLUSION

A thinner temporal RNFL and the correlation between Raynaud's phenomenon and disease duration and structural abnormalities on OCT suggest the presence of early ganglion cell damage in patients with SSc. Although mild, these findings indicate the need to monitor ocular abnormalities in SSc.

Optic Nerve Head Hemoglobin Levels in Glaucoma: A Structural and Functional Correlation Study

Purpose

To investigate structural and functional correlations in glaucoma patients using optic nerve head hemoglobin (ONH Hb) measurements as determined by automated colorimetric analysis of conventional retinography.

Methods

We prospectively enrolled healthy participants and glaucomatous patients with a wide range of disease stages. All participants underwent visual field (VF) testing (standard automated perimetry, SAP), color fundus imaging (mydriatic retinography), and peripapillary retinal nerve fiber layer (pRNFL) assessment through spectral-domain optical coherence tomography (SD-OCT). Software Laguna ONhE was used to estimate the amount of ONH Hb and to determine the glaucoma discriminant function (GDF) index. Scatter plots were constructed, and regression analysis was used to investigate the correlations between GDF, average pRNFL thickness, and VF mean deviation (VFMD) index values. A secondary analysis was performed to compare each parameter between three different glaucoma groups divided according to VFMD values (mild, >−6 dB; moderate, −6 to −12 dB; and advanced, <−12 dB).

Results

One hundred ninety-six eyes from 123 participants (69 with glaucoma and 54 controls) were enrolled. Overall, all parameters evaluated differed significantly between glaucomatous and control eyes (p ≤ 0.001). The comparison of each parameter according to groups of disease stages revealed significant differences between controls and each of the glaucomatous groups (p < 0.001). More pronounced changes in GDF values were observed in early disease stages. We found significant nonlinear correlations between GDF and VFMD values (R² = 0.295, p < 0.001) and between pRNFL thickness and VFMD (R² = 0.598, p < 0.001). A linear correlation was found between GDF and pRNFL thickness values (R² = 0.195, p < 0.001).

Conclusion

Our results showed significant associations between ONH Hb values and both structural and functional damage in glaucoma obtained by SD-OCT and SAP, respectively. The nonlinear correlation we found and the GDF behavior along different disease stages suggest that ONH Hb levels' reduction may precede visual function changes in early glaucoma stages.

Multimodal Machine Learning Using Visual Fields and Peripapillary Circular OCT Scans in Detection of Glaucomatous Optic Neuropathy

PURPOSE

To develop and validate a multimodal artificial intelligence algorithm, FusionNet, using the pattern deviation probability plots from visual field (VF) reports and circular peripapillary OCT scans to detect glaucomatous optic neuropathy (GON).

DESIGN

Cross-sectional study.

SUBJECTS

Two thousand four hundred sixty-three pairs of VF and OCT images from 1083 patients.

METHODS

FusionNet based on bimodal input of VF and OCT paired data was developed to detect GON. Visual field data were collected using the Humphrey Field Analyzer (HFA). OCT images were collected from 3 types of devices (DRI-OCT, Cirrus OCT, and Spectralis). Two thousand four hundred sixty-three pairs of VF and OCT images were divided into 4 datasets: 1567 for training (HFA and DRI-OCT), 441 for primary validation (HFA and DRI-OCT), 255 for the internal test (HFA and Cirrus OCT), and 200 for the external test set (HFA and Spectralis). GON was defined as retinal nerve fiber layer thinning with corresponding VF defects.

MAIN OUTCOME MEASURES

Diagnostic performance of FusionNet compared with that of VFNet (with VF data as input) and OCTNet (with OCT data as input).

RESULTS

FusionNet achieved an area under the receiver operating characteristic curve (AUC) of 0.950 (0.931-0.968) and outperformed VFNet (AUC, 0.868 [95% confidence interval (CI), 0.834-0.902]), OCTNet (AUC, 0.809 [95% CI, 0.768-0.850]), and 2 glaucomatologists (glaucomatologist 1: AUC, 0.882 [95% CI, 0.847-0.917]; glaucomatologist 2: AUC, 0.883 [95% CI, 0.849-0.918]) in the primary validation set. In the internal and external test sets, the performances of FusionNet were also superior to VFNet and OCTNet (FusionNet vs VFNet vs OCTNet: internal test set 0.917 vs 0.854 vs 0.811; external test set 0.873 vs 0.772 vs 0.785). No significant difference was found between the 2 glaucomatologists and FusionNet in the internal and external test sets, except for glaucomatologist 2 (AUC, 0.858 [95% CI, 0.805-0.912]) in the internal test set.

CONCLUSIONS

FusionNet, developed using paired VF and OCT data, demonstrated superior performance to both VFNet and OCTNet in detecting GON, suggesting that multimodal machine learning models are valuable in detecting GON.

Structure-Function Mapping Using a Three-Dimensional Neuroretinal Rim Parameter Derived From Spectral Domain Optical Coherence Tomography Volume Scans

Purpose

To assess the structure-function relationship in glaucoma using Humphrey visual field (HVF) perimetry and a three-dimensional neuroretinal rim parameter derived from spectral domain optical coherence tomography (SD-OCT) volume scans.

Methods

Structure-function correlation was analyzed globally and regionally (four quadrants and four sectors). Structural data included peripapillary retinal nerve fiber layer (RNFL) thickness and minimum distance band (MDB) neuroretinal rim thickness, defined as the shortest distance between the inner cup surface and the outer retinal pigment epithelium/Bruch's membrane complex. Logarithmic regression analyses were performed and Pearson correlation coefficients determined to assess relationship strength.

Results

The study consisted of 102 open-angle glaucoma patients and 58 healthy subjects. The Pearson correlation coefficient for global MDB thickness (R = 0.585) was higher than for global RNFL thickness (R = 0.492), but the difference was not statistically significant (P = 0.18). The correlation coefficients for regional MDB thicknesses and corresponding HVF sensitivities were higher than those for regional RNFL thicknesses and HVF in six out of eight regions (P = 0.08 to 0.47). In the remaining two out of eight regions, the correlation coefficients were higher for RNFL thickness than for MDB thickness (P = 0.15 to 0.20).

Conclusions

Three-dimensional MDB neuroretinal rim thickness relates to visual function as strongly as the most commonly used SD-OCT parameter for glaucoma, two-dimensional peripapillary RNFL thickness.

Translational Relevance

This paper illustrates the potential for 3D OCT algorithms to improve in vivo imaging in glaucoma.

Elucidation of the role of the lamina cribrosa in glaucoma using optical coherence tomography

Glaucoma is a chronic and progressive optic neuropathy characterized by the death of retinal ganglion cells and corresponding visual field loss. Despite the growing number of studies on the subject, the pathogenesis of the disease remains unclear. Notwithstanding, several studies have shown that the lamina cribrosa (LC) is considered an anatomic site of glaucomatous optic nerve injury, thus having a key role in the pathophysiology of glaucoma development and progression. Different morphological alterations of the LC have been described in vivo in glaucomatous eyes after the evolution of optical coherence tomography (OCT) devices. The most relevant findings were the reduction of laminar thickness, the presence of localized defects, and the posterior LC displacement. These new laminar parameters documented through OCT are not only promising as possible additional tools for glaucoma diagnosis and monitoring, but also as predictors of disease progression. In spite of the advance of technology, however, proper evaluation of the LC is not yet viable in all eyes. We describe OCT-identified LC changes related to the development and progression of glaucoma and provide future directions based on a critical data analysis, focusing on its clinical relevance and applicability.

Comparison of Rate of Change between Bruch's Membrane Opening Minimum Rim Width and Retinal Nerve Fiber Layer in Eyes Showing Optic Disc Hemorrhage

PURPOSE

To investigate and compare the longitudinal rate of change of Bruch's membrane opening minimum rim width (BMO-MRW) and peripapillary retinal nerve fiber layer (RNFL) thickness in eyes showing optic disc hemorrhage(DH).

DESIGN

Observational case series.

METHODS

A total of 82 subjects(82 eyes) showing DH who had undergone more than five reliable spectral-domain optical coherence tomography (OCT) tests were included. BMO-MRW and RNFL were measured with OCT at 3-month intervals. The rates of change in global and each Garway-Heath sector were calculated with a linear mixed-effects model after adjusting for age, sex, and BMO area.

RESULTS

The mean follow-up period was 21.57 ± 7.88 months with a mean number of 7.88 ± 2.39 OCT tests. Baseline demographics were age (58.37 ± 10.65 y); 46.3% were female; and the mean deviation was -4.41 ± 5.04 dB. The global rate of change in BMO-MRW was -3.507 ± 0.675 μm/y and in -1.404 ± 0.208 μm/y in RNFL. The rate of change was the greatest in the inferotemporal sector, which was -9.141 ± 1.254 μm/y in BMO-MRW and -4.204 ± 0.490 μm/y in the RNFL. The rate of change was significantly greater in BMO-MRW than in the RNFL in all sectors, except for the nasal sector (P < .05). Percentage of reduction was significantly greater in BMO-MRW than in RNFL in the inferotemporal and superotemporal sectors (P < .05).

CONCLUSIONS

BMO-MRW showed a significantly greater rate of change than RNFL in eyes showing DH, especially in the inferotemporal and superotemporal sectors in percentage of reduction. Thus, it may be more advantageous to detect glaucomatous progression earlier in BMO-MRW than in the RNFL in eyes showing DH that are more likely to progress.

Rate of Change in Bruch's Membrane Opening-Minimum Rim Width and Peripapillary RNFL in Early Normal Tension Glaucoma

Background: to investigate the rate of change (ROC) of Bruch’s membrane opening minimum rim width (BMO-MRW) and peripapillary retinal nerve fiber layer (RNFL) thickness in early normal tension glaucoma (NTG) patients. Methods: in this longitudinal cohort study, 115 subjects (115 eyes) diagnosed as early NTG (mean deviation > −6.0 dB) and who had completed more than five times of spectral-domain optical coherence tomography (OCT) tests with acceptable quality were included. Measurement of BMO-MRW and RNFL were performed at 3-month intervals by OCT. Linear mixed-effects model was employed to calculate the ROC in global region and six Garway-Heath sectors with adjusting age, sex, and BMO area. Results: Average follow-up was 20.99 ± 6.99 months with OCT number of 7.54 ± 2.12. Baseline intraocular pressure was 14.72 ± 2.70 mmHg and MD was −2.73 ± 2.26 dB. ROC of global BMO-MRW was −2.06 ± 0.65 µm/yr and RNFL was −0.96 ± 0.16 µm/yr (p = 0.098). The most rapid ROC was in inferotemporal sector (BMO-MRW: −3.02 ± 0.88 µm/yr, RNFL: −1.96 ± 0.36 µm/yr) followed by superotemporal sector. Conclusion: The ROC of BMO-MRW, the new parameter along with that of RNFL should be considered in the management of early NTG. BMO-MRW may show visible reduction ROC better than RNFL to detect early progression in early NTG when visual field may not show significant change.

Using Enhanced Depth Imaging Optical Coherence Tomography-Derived Parameters to Discriminate between Eyes with and without Glaucoma: A Cross-Sectional Comparative Study

INTRODUCTION

New technologies have been developed in order to decrease interpersonal influence and subjectivity during the glaucoma diagnosis process. Enhanced depth imaging spectral-domain OCT (EDI OCT) has turned up as a favorable tool for deep optic nerve head (ONH) structures assessment.

OBJECTIVE

A prospective cross-sectional study was conducted to compare the diagnostic performance of different EDI OCT-derived parameters to discriminate between eyes with and without glaucoma.

MATERIAL AND METHODS

The following ONH parameters were measured: lamina cribrosa (LC) thickness and area; prelaminar neural tissue (PLNT) thickness and area; average Bruch's membrane opening - minimum rim width (BMO-MRW), superior BMO-MRW, and inferior BMO-MRW. Peripapillary retinal nerve fiber layer (pRNFL) thickness was also obtained.

RESULTS

Seventy-three participants were included. There were no significant differences between AUCs for average BMO-MRW (0.995), PLNT area (0.968), and average pRNFL thickness (0.975; p ≥ 0.089). However, AUCs for each of these 3 parameters were significantly larger than LC area AUC (0.701; p ≤ 0.001). Sensitivities at 80% specificity were: PLNT area = 92.3%, average BMO-MRW = 97.4%, and average pRNFL thickness = 94.9%.

CONCLUSIONS

Comparing the diagnostic performance of different EDI OCT ONH parameters to discriminate between eyes with and without glaucoma, we found better results for neural tissue-based indexes (BMO-MRW and PLNT area) compared to laminar parameters. In this specific population, these neural tissue-based parameters (including PLNT area, which was investigated by the first time in the present study) had a diagnostic performance comparable to that of the conventional pRNFL thickness protocol.

Deep Learning Approaches Predict Glaucomatous Visual Field Damage from OCT Optic Nerve Head En Face Images and Retinal Nerve Fiber Layer Thickness Maps

PURPOSE

To develop and evaluate a deep learning system for differentiating between eyes with and without glaucomatous visual field damage (GVFD) and predicting the severity of GFVD from spectral domain OCT (SD OCT) optic nerve head images.

DESIGN

Evaluation of a diagnostic technology.

PARTICIPANTS

A total of 9765 visual field (VF) SD OCT pairs collected from 1194 participants with and without GVFD (1909 eyes).

METHODS

Deep learning models were trained to use SD OCT retinal nerve fiber layer (RNFL) thickness maps, RNFL en face images, and confocal scanning laser ophthalmoscopy (CSLO) images to identify eyes with GVFD and predict quantitative VF mean deviation (MD), pattern standard deviation (PSD), and mean VF sectoral pattern deviation (PD) from SD OCT data.

MAIN OUTCOME MEASURES

Deep learning models were compared with mean RNFL thickness for identifying GVFD using area under the curve (AUC), sensitivity, and specificity. For predicting MD, PSD, and mean sectoral PD, models were evaluated using R² and mean absolute error (MAE).

RESULTS

In the independent test dataset, the deep learning models based on RNFL en face images achieved an AUC of 0.88 for identifying eyes with GVFD and 0.82 for detecting mild GVFD significantly (P < 0.001) better than using mean RNFL thickness measurements (AUC = 0.82 and 0.73, respectively). Deep learning models outperformed standard RNFL thickness measurements in predicting all quantitative VF metrics. In predicting MD, deep learning models based on RNFL en face images achieved an R² of 0.70 and MAE of 2.5 decibels (dB) compared with 0.45 and 3.7 dB for RNFL thickness measurements. In predicting mean VF sectoral PD, deep learning models achieved high accuracy in the inferior nasal (R² = 0.60) and superior nasal (R² = 0.67) sectors, moderate accuracy in inferior (R² = 0.26) and superior (R² = 0.35) sectors, and lower accuracy in the central (R² = 0.15) and temporal (R² = 0.12) sectors.

CONCLUSIONS

Deep learning models had high accuracy in identifying eyes with GFVD and predicting the severity of functional loss from SD OCT images. Accurately predicting the severity of GFVD from SD OCT imaging can help clinicians more effectively individualize the frequency of VF testing to the individual patient.

Characteristics of Patients Showing Discrepancy Between Bruch's Membrane Opening-Minimum Rim Width and Peripapillary Retinal Nerve Fiber Layer Thickness

Background: To investigate clinical characteristics of patients showing discrepancy between Bruch’s membrane opening minimum rim width (BMO-MRW) and peripapillary retinal nerve fiber layer (RNFL) thickness. Correlation with the visual field (VF) was also inspected. Methods: In this prospective, cross-sectional study, 106 eyes (106 subjects) showing normal BMO-MRW classification but abnormal RNFL classification were included. All patients underwent confocal scanning laser ophthalmoscopy, spectral-domain optical coherence tomography, and standard automated perimetry. Results: Clinical characteristics were as follows: mean age: 52.79 ± 14.75 years; spherical equivalent (SE), −2.52 ± 3.48 diopter (D); SE < −5.0 D, 34 (32.1%) eyes; large disc (>2.43 mm²), 40.6%; small disc (<1.63 mm²), 12.5%; VF index, 96.72 ± 9.58%; mean deviation, −1.74 ± 3.61 dB; β-peripapillary atrophy (PPA), 96.2%; γ-PPA, 75.5%. Majority (86.1%) of these cases demonstrated normal (71.3%) or borderline (14.9%) on VF. Temporal and nasal RNFL showed significant differences among disc size subgroups (all p < 0.05). Nasal RNFL was significantly thicker in a large disc group than other subgroups. Temporal, superotemporal, inferotemporal, inferonasal RNFL, and superior RNFL peak location showed significant differences (all p < 0.05) among SE subgroups. Temporal RNFL was significantly thicker in the high myopia group than other subgroups. Conclusions: Temporalization of RNFL peaks in myopia and nasalization of RNFL peaks in large disc that display abnormal classifications might show normal classification of BMO-MRW. These findings of discrepancy between classifications should be considered in the diagnosis of early glaucoma.