Abstract
Purpose :
Primary care clinic based diabetic retinopathy (DR) screening is a potential solution to improve the low national DR screening rates, but is limited by the high cost, bulky nature, and specialized training required of traditional table-top retinal cameras. Smartphone-based retinal photography can help address these limitations. The goal of this study is to improve the usability of a smartphone-based retinal camera, CellScope Retina, among medical staff who are inexperienced with retinal imaging.
Methods :
24 medical assistants and technicians were recruited for a total of 4 rounds of usability testing. Participants were given a 1-minute tutorial on how the smartphone-based retinal camera worked, and then asked to capture photos from 5 fields of the retina of a model eye. The duration of image acquisition was documented. Software, hardware, and instructional modifications were made after each round of testing in accordance with user feedback. Afterwards, a proof-of-concept test was performed on the dilated eye of a human volunteer. An IRB exemption for this study was granted due to no direct patient involvement.
Results :
There was an overall decrease in average image capture time after each round (Trial 1: 260 ± 60 seconds for 1 image, 325 ± 60 seconds for 5 images; Trial 2: 55 ± 20 seconds for 1 image, 121 ± 41 seconds for 5 images; Trial 3: 43 ± 16 seconds for 1 image, 108 ± 13 for 5 images; Trial 4: 34 ± 17 seconds for 1 image, 119 ± 26 seconds for 5 images; Human trial: 66 ± 7 seconds for 1 image, 229 ± 114 for 5 images).
Conclusions :
CellScope Retina allows medical assistants who are naïve to retinal photography to rapidly (in < 1 minute) acquire a high quality photograph of the retina. Afterward, a test performed on a human eye demonstrated clear images with only a small increase in imaging time. Usability testing is a rapid, high-yield approach for feedback-driven improvements of smartphone-based retinal photography among inexperienced users. Additional testing with human subjects is needed for further improvements.
This is an abstract that was submitted for the 2017 ARVO Annual Meeting, held in Baltimore, MD, May 7-11, 2017.