Detecting Progression in Advanced Glaucoma: Are Optical Coherence Tomography Global Metrics Viable Measures?

SIGNIFICANCE: Optical coherence tomography (OCT) summary measures have been suggested as a way to detect progression in eyes with advanced glaucoma. Here, we show that these measures have serious flaws largely due to segmentation errors. However, inspection of the images and thickness maps can be clinically useful. PURPOSE: This study aimed to test the hypothesis that recently suggested global OCT measures for detecting progression in eyes with advanced progression are seriously affected by segmentation mistakes and other errors that limit their clinical utility. METHODS: Forty-five eyes of 38 patients with a 24-2 mean deviation worse than -12 dB had at least two spectral domain OCT sessions (0.8 to 4.4 years apart) with 3.5-mm circle scans of the disc and cube scans centered on the fovea. Average (global) circumpapillary retinal nerve fiber layer thickness, GcRNFL, and ganglion cell plus inner plexiform layer thickness, GGCLP, were obtained from the circle and cube scan, respectively. To evaluate progression, ΔGcRNFL was calculated for each eye as the GcRNFL value at time 2 minus the value at time 1, and ΔGGCLP was calculated in a similar manner. The b-scans of the six eyes with the highest and lowest ΔGcRNFL and ΔGGCLP values were examined for progression as well as segmentation, alignment, and centering errors. RESULTS: Progression was a major factor in only 7 of the 12 eyes with the most negative values of either ΔGcRNFL or ΔGGCLP, whereas segmentation played a role in 8 eyes and was the major factor in all 12 eyes with the largest positive values. In addition, alignment (one eye) and other (three eyes) errors played a secondary role in four of the six eyes with the most negative ΔGcRNFL values. CONCLUSIONS: For detecting the progression of advanced glaucoma, common summary metrics have serious flaws largely due to segmentation errors, which limit their utility in clinical and research settings.

Distinguishing Healthy From Glaucomatous Eyes With Optical Coherence Tomography Global Circumpapillary Retinal Nerve Fiber Thickness in the Bottom 5th Percentile.

PRCIS: Two novel, quantitative metrics, and 1 traditional metric were able to distinguish between many, but not all healthy and glaucomatous eyes in the bottom 5th percentile of global circumpapillary retinal nerve fiber layer (cpRNFL) thickness. PURPOSE: To test the hypothesis that objective optical coherence tomography measures can distinguish between a healthy control with global cpRNFL thickness within the lower 5% of normal and a glaucoma patient with an equivalent cpRNFL thickness. PATIENTS AND METHODS: A total of 37 healthy eyes from over 700 normative eyes fell within the bottom 5th percentile in global cpRNFL thickness. The global cpRNFL thickness of 35 glaucomatous eyes from 188 patients fell within the same range. For the traditional methods, the global cpRNFL thickness percentile and the global ganglion cell layer (GCL) thickness percentile for the central ±8 degrees, were calculated for all 72 eyes. For the novel cpRNFL method, the normalized root mean square (RMS) difference between the cpRNFL thickness profile and the global thickness-matched normative thickness profile was calculated. For the superior-inferior (SI) GCL method, the normalized mean difference in superior and inferior GCL thickness was calculated for the central ±8 degrees. RESULTS: The best quantitative metric, the RMS cpRNFL method, had an accuracy of 90% compared with 81% for the SI GCL and 81% for the global GCL methods. As expected, the global cpRNFL had the worst accuracy, 72%. Similarly, the RMS cpRNFL method had an area under the curve of 0.93 compared with 0.83 and 0.84 for the SI GCL and global GCL methods, respectively. The global cpRNFL method had the worst area under the curve, 0.75. CONCLUSION: Quantitative metrics can distinguish between most of the healthy and glaucomatous eyes with low global cpRNFL thickness. However, even the most successful metric, RMS cpRNFL, missed some glaucomatous eyes.

The OCT RNFL Probability Map and Artifacts Resembling Glaucomatous Damage.

Purpose: The purpose of this study was to improve the diagnostic ability of the optical coherence tomography (OCT) retinal nerve fiber layer (RNFL) probability (p-) map by understanding the frequency and pattern of artifacts seen on the p-maps of healthy control (HC) eyes resembling glaucomatous damage. Methods: RNFL p-maps were generated from wide-field OCT cube scans of 2 groups of HC eyes, 200 from a commercial normative group (HC-norm) and 54 from a prospective study group, as well as from 62 patient eyes, which included 32 with early glaucoma (EG). These 32 EG eyes had 24-2 mean deviation (MD) better than -6 dB and perimetric glaucoma as defined by 24-2 and 10-2 criteria. For the HC groups, "glaucoma-like" arcuates were defined as any red region near the temporal half of the disc. Results: Seven percent of the 200 HC-norm and 11% of the 54 HC RNFL p-maps satisfied the definition of "glaucoma-like," as did all the patients' p-maps. The HC p-maps showed two general patterns of abnormal regions, "arcuate" and "temporal quadrant," and these patterns resembled those seen on some of the RNFL p-maps of the EG eyes. A "vertical midline" rule, which required the abnormal region to cross the vertical midline through the fovea, had a specificity of >99%, and a sensitivity of 75% for EG and 93% for moderate to advanced eyes. Conclusions: Glaucoma-like artifacts on RNFL p-maps are relatively common and can masquerade as arcuate and/or widespread/temporal damage. Translational Relevance: A vertical midline rule had excellent specificity. However, other OCT information is necessary to obtain high sensitivity, especially in eyes with early glaucoma.

Strategies to Improve Convolutional Neural Network Generalizability and Reference Standards for Glaucoma Detection From OCT Scans.

Purpose: To develop and evaluate methods to improve the generalizability of convolutional neural networks (CNNs) trained to detect glaucoma from optical coherence tomography retinal nerve fiber layer probability maps, as well as optical coherence tomography circumpapillary disc (circle) b-scans, and to explore impact of reference standard (RS) on CNN accuracy. Methods: CNNs previously optimized for glaucoma detection from retinal nerve fiber layer probability maps, and newly developed CNNs adapted for glaucoma detection from optical coherence tomography b-scans, were evaluated on an unseen dataset (i.e., data collected at a different site). Multiple techniques were used to enhance CNN generalizability, including augmenting the training dataset, using multimodal input, and training with confidently rated images. Model performance was evaluated with different RS. Results: Training with data augmentation and training on confident images enhanced the accuracy of the CNNs for glaucoma detection on a new dataset by 5% to 9%. CNN performance was optimal when a similar RS was used to establish labels both for the training and the testing sets. However, interestingly, the CNNs described here were robust to variation in the RS. Conclusions: CNN generalizability can be improved with data augmentation, multiple input image modalities, and training on images with confident ratings. CNNs trained and tested with the same RS achieved best accuracy, suggesting that choosing a thorough and consistent RS for training and testing improves generalization to new datasets. Translational Relevance: Strategies for enhancing CNN generalizability and for choosing optimal RS should be standard practice for CNNs before their deployment for glaucoma detection.

Improving the Detection of Glaucoma and Its Progression: A Topographical Approach.

Glaucoma is typically defined as a progressive optic neuropathy characterized by a specific (arcuate) pattern of visual field (VF) and anatomic changes. Therefore, we should be comparing arcuate patterns of damage seen on VFs with those seen on optical coherence tomography (OCT) maps. Instead, clinicians often use summary metrics such as VF pattern standard deviation, OCT retinal nerve fiber (RNF) global thickness, etc. There are 2 major impediments to topographically comparing patterns of damage on VF and OCT maps. First, until recently, it was not easy to make these comparisons with commercial reports. While recent reports do make it easier to compare VF and OCT maps, they have shortcomings. In particular, the 24-2 VF covers a larger retinal region than the commercial OCT scans, and, further, it is not easy to understand the topographical relationship among the different maps/plots within the current OCT reports. Here we show how a model of RNF bundles can overcome these problems. The second major impediment is the lack of a quantitative, and automated, method for comparing patterns of damage seen on VF and OCT maps. However, it is now possible to objectively and automatically quantify this agreement. Together, the RNF bundle model and the automated structure-function method should improve the power of topographical methods for detecting glaucoma and its progression. This should prove useful in clinical studies and trials, as well as for training and validating artificial intelligence/deep learning approaches for these purposes.

Optical Coherence Tomography Can Be Used to Assess Glaucomatous Optic Nerve Damage in Most Eyes With High Myopia.

PRECIS: It is generally assumed that optical coherence tomography (OCT) cannot be used to diagnose glaucomatous optic neuropathy (GON) in high myopes. However, this study presents evidence that there is sufficient information in OCT scans to allow for accurate diagnosis of GON in most eyes with high myopia. PURPOSE: The purpose of this study was to test the hypothesis that glaucomatous damage can be accurately diagnosed in most high myopes via an assessment of the OCT results. PATIENTS AND METHODS: One hundred eyes from 60 glaucoma patients or suspects, referred for OCT scans and evaluation, had corrected spherical refractive errors worse than -6 D and/or axial lengths ≥26.5 mm. An OCT specialist judged whether the eye had GON, based upon OCT circle scans of the disc and cube scans centered on the macula. A glaucoma specialist made the same judgement using all available information (eg, family history, repeat visits, intraocular pressure, 10-2 and 24-2 visual fields, OCT). A reference standard was created based upon the glaucoma specialist's classifications. In addition, the glaucoma specialist judged whether the eyes had peripapillary atrophy (PPA), epiretinal membrane (ERM), tilted disc (TD), and/or a paravascular inner retinal defect (PIRD). RESULTS: The OCT specialist correctly identified 97 of the 100 eyes using the OCT information. In 63% of the cases, the inner circle scan alone was sufficient. For the rest, additional scans were requested. In addition, 81% of the total eyes had: PPA (79%), ERM (18%), PIRD (26%), and/or TD (48%). CONCLUSIONS: For most eyes with high myopia, there is sufficient information in OCT scans to allow for accurate diagnosis of GON. However, the optimal use of the OCT will depend upon training to read OCT scans, which includes taking into consideration myopia related OCT artifacts and segmentation errors, as well as PPA, ERM, PIRD, and TD.

A Topographic Comparison of OCT Minimum Rim Width (BMO-MRW) and Circumpapillary Retinal Nerve Fiber Layer (cRNFL) Thickness Measures in Eyes With or Suspected Glaucoma.

PRéCIS:: Bruch's membrane opening-minimum rim width (BMO-MRW) and circumpapillary retinal nerve fiber layer (cRNFL) thickness measures may be improved by comparing probability levels and accounting for blood vessel locations. PURPOSE: To understand the differences between 2 optical coherence tomography measures of glaucomatous damage: the BMO-MRW and cRNFL thickness. MATERIALS AND METHODS: Optical coherence tomography circle scans were obtained for an early glaucoma group (EG) of 88 eyes (88 patients) with 24-2 mean deviation better than -6.0 dB, and a broader group (BG) of 188 eyes (110 patients) with 24-2 mean deviation from -0.15 to -27.0 dB. On the basis of a commercial report, the cRNFL and BMO-MRW of each hemidisc was classified as abnormal if either of the 2 superior (inferior) sectors, temporal superior and nasal superior (temporal inferior and nasal inferior), was yellow or red (P<5%); and as normal if both were green (P≥5%). In addition, a post hoc analysis identified the reasons for disagreements on the basis of the presence (or absence) of glaucomatous damage at a hemidisc level (consensus of 4 experts). RESULTS: The BMO-MRW and cRNFL measures agreed in 81.9% (broader group) and 73.9% (EG) of the hemidiscs. In both groups, an abnormal-BMO-MRW/normal-cRNFL disagreement was as common as a normal-BMO-MRW/abnormal-cRNFL. Of the 46 EG hemidisc disagreements, the number of "mistakes" for BMO-MRW (28) was nonsignificantly higher than for cRNFL (18) (P=0.15). Primary causes for disagreement were as follows: borderline significance level, a local defect, and aberrant blood vessel location. CONCLUSIONS: Although BMO-MRW and cRNFL measures agreed in the majority of hemidiscs, they still disagreed in over 25% of the EG hemidiscs. These measures may be improved by comparing actual probability levels and accounting for blood vessel locations. However, both can miss information available on retinal ganglion cell/retinal nerve fiber layer probability maps.

Reasons why OCT Global Circumpapillary Retinal Nerve Fiber Layer Thickness is a Poor Measure of Glaucomatous Progression.

Purpose: To assess the effects of local defects, segmentation errors, and improper image alignment on the performance of the commonly used optical coherence tomography (OCT) measure of progression, that is the change in global (average) circumpapillary retinal nerve fiber layer (cpRNFL) thickness (ΔG). Methods: One hundred fifty eyes suspected of, or with, early glaucoma had OCT circle and cube scans obtained using eye tracking on two occasions at least 1 year apart. Statistical progression was defined by fixed values of ΔG (3-8 um) and quantile regression. For a reference standard, four authors identified 30 eyes as "likely progressed," and 61 eyes that "likely had not progressed" based on OCT reports from both baseline and follow-up tests. Results: A ΔG criterion of 4 um had the best accuracy: 77%, with 5 false positive (8.2%) and 16 false negative (53%). A post hoc analysis of circular b-scans and OCT probability maps of these eyes indicated that segmentation errors and local progression accounted for most of these mistakes. Segmentation errors, although less common, were also present in true positives and true negatives. Conclusions: Local defects and segmentation errors are the primary reasons for the poor performance of cpRNFL thickness G metric. Because these problems are difficult, if not impossible, to eliminate, the G metric should not be relied on in isolation for detecting glaucomatous progression. Translational Relevance: Local defects and segmentation errors are easily identified by viewing OCT circumpapillary images, which should be part of the standard protocol for detecting glaucomatous progression.

Qualitative evaluation of neuroretinal rim and retinal nerve fibre layer on optical coherence tomography to detect glaucomatous damage.

PURPOSE: To understand the added value of Bruch's membrane opening-minimum rim width (BMO-MRW) measurements to conventional circumpapillary retinal nerve fibre layer (cpRNFL) thickness measurements on optical coherence tomography (OCT) imaging for discriminating between perimetric glaucoma and healthy eyes, evaluated through a qualitative evaluation. METHODS: 384 healthy eyes and 188 glaucoma eyes were evaluated, and glaucoma eyes were categorised as perimetric (n=107) based on a history of ≥3 consecutive abnormal 24-2 visual field tests or suspected glaucoma if they did not (n=81). OCT-derived BMO-MRW and cpRNFL reports were qualitatively evaluated by two experienced graders in isolation at first, and then by using both reports combined. The diagnostic performance (sensitivity at 95% specificity, total and partial area under the receiver operating characteristic curve) of detecting perimetric glaucoma with each method were compared. RESULTS: All diagnostic performance measures for detecting perimetric glaucoma eyes were not significantly different when using either the cpRNFL or BMO-MRW reports alone compared with using both reports combined (p≥0.190), nor when comparing the use of each report in isolation (p≥0.500). CONCLUSIONS: Experienced graders exhibited no difference in discriminating between perimetric glaucoma and healthy eyes when using a cpRNFL report alone, the BMO-MRW report alone or the two reports combined. Therefore, either OCT imaging report of the neuroretinal tissue could be used effectively for detecting perimetric glaucoma, but further studies are needed to determine whether there are specific advantages of each method, or the combination of both, when evaluating eyes that have a greater degree of diagnostic uncertainty.