Face perception - can it be improved in age-related macular degeneration and Stargardt disease?

Abstract

Purpose : Face recognition difficulties are well described for people with central vision loss. With the advent of wearable displays these difficulties can potentially be aided with real-time image enhancement techniques. The aim of this study is to measure the impact of a new image enhancement technique on face perception for people with central vision loss.

Methods : We developed a face perception test, based on the Caledonian Face Test [Logan 2016], where participants are presented with four faces (3 identical distractors and 1 target) and are asked to spot the odd one out. Difficulty was measured by the distance of the target to distractor in log units. To measure the reliability of the test, 67 normally sighted participants performed the test twice, separated by a 5-minute break.
The test was then used to measure the impact of a new method, cartoon image enhancement. 19 Age-related Macular Degeneration (VA:0.79±0.4 logMAR, Age:78±8) and 11 Stargardt (VA:0.96±0.42 logMAR, Age:40±13) patients were recruited from the Oxford Eye Hospital. Participants were asked to perform the test twice once with un-enhanced faces, and once with image enhancement.

Results : Normally sighted: Average score 2.97±0.12 log units. The Coefficient of reproducibility between test 1 and 2 was 0.15 log units.
Patients: Average un-enhanced scores for AMD and Stargardt; 3.83±0.74 and 3.36±0.32.
The average improvement in face perception performance for the AMD group was 0.3±0.5 and for the Stargardt 0.024±0.16.
Patients with good contrast sensitivity (1 or 2 cycles per degree) performed the face perception test better at baseline (p<0.001,R²=0.6 and p=0.009,R²=0.39 respectively). Cartoon image enhancement increased the test scores significantly for AMD patients. In AMD patients contrast sensitivity deficit was a good predictor of the positive effect of image enhancement on their performance (p<0.001,R²=0.75). Stargardt patients however benefited more from image enhancement if they have a lower VA (p=0.004,R²=0.83).

Conclusions : We have presented a novel image enhance technique that has shown the potential to improve the visibility of faces for people with AMD. However, little improvement is shown for Stargardt patients, implying that they are preforming near their maximum potential. The cartoon enhancement approach is simple, and can easily be implemented on a wearable mobile display.

This abstract was presented at the 2019 ARVO Annual Meeting, held in Vancouver, Canada, April 28 - May 2, 2019.