Abstract
Purpose :
The Realwear HMT-1™ is a head mounted tablet device with a camera and a micro-display in use primarily for industry applications. The device may be useful for people with low vision with modification of some its features. The purpose of this study was to modify the HMT-1 to provide better low vision functionality, pilot test the modified device for reading with low vision, and test the usefulness of addition of real time kinematic (RTK) device data on improving indoor location accuracy for future navigation uses.
Methods :
The HMT-1 consists of a video display mounted to an adjustable arm and a camera with magnification and image enhancing capability. A custom Android application (Industrial Badger App™) was used to increase magnification capability and allow for text-to-speech conversion of printed reading material. Visual acuity and contrast sensitivity were measured with habitual correction and with the HMT-1. A previously validated passage reading test (300 words per passage with 3 comprehension questions per passage) was used to assess speed and comprehension using: habitual aids, HMT-1 with magnification, and text-to-speech. Device location accuracy was tested at 8 indoor locations for the stock HMT-1 location provider and with the addition of RTK data, and GeoJSON centroids were calculated to compare location error between conditions.
Results :
Three participants with low vision were recruited for pilot reading testing. Mean±SD age was 64±5 and 66% were male. Mean better-eye baseline ETDRS letter VA was 45±12 (approximately 20/120) and improved to 78±11 (approximately 20/25, (p<0.001). Mean passage reading time with habitual aid was 183 seconds vs. 457 seconds with HMT-1 camera (one subject was unable to use the camera due to poor VA). Comprehension performance was worse with HMT-1 camera compared to both habitual aid and HMT text to speech. All subjects reported the device to be useful and 2/3 of subjects reported they would use HMT-1 in real life. The addition of RTK data resulted in reduction of the average error of the cluster centroid from 60m to 38.4m and generally faster responses.
Conclusions :
A modified version of the HMT-1 showed promise as an aid for near tasks in people with low vision, especially with reading and indoor navigation. Future work will seek to improve reading speed and location accuracy for navigation.
This is a 2021 ARVO Annual Meeting abstract.