Comparison of Visual Field Test Measurements With a Novel Approach on a Wearable Headset to Standard Automated Perimetry

Abstract

Prcis: This study of inter-test comparability of a novel visual field application installed on an augmented-reality portable headset and Humphrey field analyzer Swedish interactive thresholding algorithm (SITA) Standard visual field test demonstrates the excellent correlation of mean deviation (MD) and mean sensitivity (MS).

Purpose: To determine the correlation between visual field testing with novel software on a wearable headset versus standard automated perimetry.

Patients and methods: Patients with and without visual field defects attributable to glaucoma had visual field testing in one eye of each patient with 2 methods: re:Imagine Strategy (Heru, Inc.) and the Humphrey field analyzer (Carl Zeiss Meditec, Inc.) SITA Standard 24-2 program. Main outcome measures included MS and MD, which were evaluated by linear regression, intraclass correlation coefficient (ICC), and Bland Altman analysis for assessment of the mean difference and limits of agreement.

Results: Measurements from 89 eyes of 89 patients (18 normal and 71 glaucomas) were compared with both instruments. Linear regression analysis demonstrated an excellent Pearson correlation coefficient of r = 0.94 for MS and r = 0.95 for MD. ICC analysis demonstrated high levels of concordance (ICC = 0.95, P < 0.001 for MS and ICC = 0.94, P < 0.001 for MD). Bland-Altman analysis determined a small mean difference between the two devices (Heru minus Humphrey) of 1.15 dB for MS and 1.06 dB for MD.

Conclusions: The Heru visual field test correlated well with SITA Standard in a population of normal eyes and eyes with glaucoma.

FIGURE 1

A patient wearing the HMD with the light shield mounted to prevent light leakage.

FIGURE 2

Plots showing the correlations between HFA and Heru of MS and MD. A, Linear regression plot of Heru MS versus HFA MS (in dB). B, Bland-Altman plot showing the difference in MS between devices as a function of the average sensitivity (in dB). C, Linear regression plot of Heru MD versus HFA MD (in dB). D, Bland-Altman plot showing differences between Heru MD and HFA MD (in dB). The 2 vertical orange lines stratify by the severity of disease based on HAP criteria (Left: mild or no disease (MD ≥ −6 dB); middle: moderate disease (−6 dB > MD ≥ −12 dB); right: severe disease (MD < −12 dB). HFA indicates Humphrey field analyzer; MD, mean deviation; MS, mean sensitivity.

FIGURE 3

Four examples of Heru tests compared with HFA tests from the same eye. Top left (ID 860): 61-year-old man with pigmentary glaucoma—on 3 glaucoma medications with a C/D of 0.7 and mean RNFL of 55 µ. Top right (ID 876): 63-year-old man with uveitic glaucoma—on 1 glaucoma medication with a C/D of 0.9 and mean RNFL of 81 µ. Bottom left (ID 870): 59-year-old man with anatomic narrow-angle—on no glaucoma medications without glaucomatous optic neuropathy. Bottom right (ID 930): 80-year-old man with normal tension glaucoma—on 4 glaucoma medications with a C/D of 0.95 and mean RNFL of 64 µ. C/D indicates cup-to-disc; HFA, Humphrey field analyzer; RNFL, retinal nerve fiber layer.

FIGURE 4

Linear regression plot of 4628 individual test points (Heru vs HFA). The 2 points contiguous with the blind spot have been omitted for both tests. Points in the plot may represent more than one individual test point. HFA indicates Humphrey field analyzer; SITA, Swedish interactive thresholding algorithm.

FIGURE 5

Pearson correlation coefficient (r) between Heru and HFA tests for nonblind spot pointwise threshold values: (A) for all participants, (B) for healthy eyes, (C) for mild eyes, (D) for moderate eyes, and (E) for severe eyes. Correlation values of <0.40 are highlighted in red. HFA indicates Humphrey field analyzer.