Do Additional Testing Locations Improve the Detection of Macular Perimetric Defects in Glaucoma?

Spatial Summation in the Glaucomatous Macula: A Link With Retinal Ganglion Cell Damage

Purpose

The purpose of this study was to test whether functional loss in the glaucomatous macula is characterized by an enlargement of Ricco's area (RA) through the application of a computational model linking retinal ganglion cell (RGC) damage to perimetric sensitivity.

Methods

One eye from each of 29 visually healthy subjects <40 years old, 30 patients with glaucoma, and 20 age-similar controls was tested with a 10-2 grid with stimuli of 5 different area sizes. Structural estimates of point-wise RGC density were obtained from optical coherence tomography (OCT) scans. Structural and functional data from the young healthy cohort were used to estimate the parameters of a computational spatial summation model to generate a template. The template was fitted with a Bayesian hierarchical model to estimate the latent RGC density in patients with glaucoma and age-matched controls. We tested two alternative hypotheses: fitting the data by translating the template horizontally (H1: change in RA) or vertically (H2: loss of sensitivity without a change in RA). Root mean squared error (RMSE) of the model fits to perimetric sensitivity were compared. Ninety-five percent confidence intervals were bootstrapped. The dynamic range of the functional and structural RGC density estimates was denoted by their 1st and 99th percentiles.

Results

The RMSE was 2.09 (95% CI = 1.92-2.26) under H1 and 2.49 (95% CI = 2.24-2.72) under H2 (P < 0.001). The average dynamic range for the structural RGC density estimates was only 11% that of the functional estimates.

Conclusions

Macular sensitivity loss in glaucoma is better described by a model in which RA changes with RGC loss. Structural measurements have limited dynamic range.

Improving the Accuracy and Speed of Visual Field Testing in Glaucoma With Structural Information and Deep Learning

Purpose

To assess the performance of a perimetric strategy using structure-function predictions from a deep learning (DL) model.

Methods

Visual field test-retest data from 146 eyes (75 patients) with glaucoma with (median [5th-95th percentile]) 10 [7, 10] tests per eye were used. Structure-function predictions were generated with a previously described DL model using cicumpapillary optical coherence tomography (OCT) scans. Structurally informed prior distributions were built grouping the observed measured sensitivities for each predicted value and recalculated for each subject with a leave-one-out approach. A zippy estimation by sequential testing (ZEST) strategy was used for the simulations (1000 per eye). Ground-truth sensitivities for each eye were the medians of the test-retest values. Two variations of ZEST were compared in terms of speed (average total number of presentations [NP] per eye) and accuracy (average mean absolute error [MAE] per eye), using either a combination of normal and abnormal thresholds (ZEST) or the calculated structural distributions (S-ZEST) as prior information. Two additional versions of these strategies employing spatial correlations were tested.

Results

S-ZEST was significantly faster, with a mean average NP of 213.87 (SD = 28.18), than ZEST, with a mean average NP of 255.65 (SD = 50.27) (P < 0.001). The average MAE was smaller for S-ZEST (1.98; SD = 2.37) than ZEST (2.43; SD = 2.69) (P < 0.001). Spatial correlations further improved both strategies (P < 0.001), but the differences between ZEST and S-ZEST remained significant (P < 0.001).

Conclusions

DL structure-function predictions can significantly improve perimetric tests.

Translational Relevance

DL structure-function predictions from clinically available OCT scans can improve perimetry in glaucoma patients.

Comparison of Compass Suprathreshold Screening Strategies

Screening programs may be useful to reduce the rate of undetected glaucoma. Compass (CMP, CenterVue, Padova, Italy) Standard Suprathreshold strategy (SST) analyses the visual function at 52 retinal locations. A new Quick Suprathreshold strategy (QST) reduces the number of tested locations to 24. With both tests, the CMP also provides an image of the central retina and a detail of the optic nerve head. The aim of this paper is to measure the performances of SST and QST compared with clinical diagnosis. 63 consecutive healthy subjects and 60 consecutive patients with perimetric defects from glaucoma in both eyes were recruited. They received one test per eye (SST or QST in randomized order); results were classified into normal, suspect and abnormal according to a global index provided in the report and called SupraThreshold Response (STR). Agreement between clinical diagnosis and test output were calculated, and test time was also measured. The agreement with the clinical diagnosis was 95.7% for SST and 95.1% for QST. When two suspect tests were excluded, agreement for QST increased to 96.7%. Test duration was 164 ± 82 s for SST and 71 ± 41 s for QST (p < 0.0001). Such a difference was similar in both glaucoma patients (respectively 231 ± 65 s vs. 105 ± 33 s, p < 0.0001) and normal subjects (98 ± 17 and 39 ± 9 s, p < 0.0001). In conclusion, SST and QST showed similar, high agreement with clinical judgement. Morphological analysis is potentially helpful in further improving the clinical usefulness of suprathreshold tests. QST is characterized by a strong time saving compared with SST.