Word recognition: re-thinking prosthetic vision evaluation

Testing Vision Is Not Testing For Vision

Visual prostheses aim to restore, at least to some extent, vision that leads to the type of perception available for sighted patients. Their effectiveness is almost always evaluated using clinical tests of vision. Clinical vision tests are designed to measure the limits of parameters of a functioning visual system. I argue here that these tests are rarely suited to determine the ability of prosthetic devices and other therapies to restore vision. This paper describes and explains many limitations of these evaluations. Prosthetic vision testing often makes use of multiple-alternative forced-choice (MAFC) procedures. Although these paradigms are suitable for many studies, they are frequently problematic in vision restoration evaluation. Two main types of problems are identified: (1) where nuisance variables provide spurious cues that can be learned in repeated training, which is common in prosthetic vision, and thus defeat the purpose of the test; and (2) even though a test is properly designed and performed, it may not actually measure what the researchers believe, and thus the interpretation of results is wrong. Examples for both types of problems are presented. Additional problems arise from confounding factors in the administration of tests are pointed as limitations of current device evaluation. For example, head tracing of magnified objects enlarged to compensate for the system's low resolution, in distinction from the scanning head (camera) movements with which users of prosthetic devices expand the limited field of view. Because of these problems, the ability to perform satisfactorily on the clinical tests is necessary but insufficient to prove vision restoration, therefore, additional tests are needed. I propose some directions to pursue in such testing.

Translational Relevance

Numerous prosthetic devices are being developed and introduced to the market. Proving the utility of these devices is crucial for regulatory and even for post market acceptance, which so far has largely failed, in my opinion. Potential reasons for the failures despite success in regulatory testing and directions for designing improved testing are provided. It is hoped that improved testing will guide improved designs of future prosthetic systems and other vision restoration approaches.

Motion Parallax Improves Object Recognition in the Presence of Clutter in Simulated Prosthetic Vision

Purpose

Efficacy of current visual prostheses in object recognition is limited. Among various limitations to be addressed, such as low resolution and low dynamic range, here we focus on reducing the impact of background clutter on object recognition. We have proposed the use of motion parallax via head-mounted camera lateral scanning and computationally stabilizing the object of interest (OI) to support neural background decluttering. Simulations in head-mounted displays (HMD), mimicking the proposed effect, were used to test object recognition in normally sighted subjects.

Methods

Images (24° field of view) were captured from multiple viewpoints and presented at a low resolution (20 × 20). All viewpoints were centered on the OI. Experimental conditions (2 × 3) included clutter (with or without) × head scanning (single viewpoint, 9 coherent viewpoints corresponding to subjects' head positions, and 9 randomly associated viewpoints). Subjects used lateral head movements to view OIs in the HMD. Each object was displayed only once for each subject.

Results

The median recognition rate without clutter was 40% for all head scanning conditions. Performance with synthetic background clutter dropped to 10% in the static condition, but it was improved to 20% with the coherent and random head scanning (corrected P = 0.005 and P = 0.049, respectively).

Conclusions

Background decluttering using motion parallax cues but not the coherent multiple views of the OI improved object recognition in low-resolution images. The improvement did not fully eliminate the impact of background.

Translational Relevance

Motion parallax is an effective but incomplete decluttering solution for object recognition with visual prostheses.