We found a match
Your institution may have access to this item. Find your institution then sign in to continue.
- Title
A pilot autonomous AI‐based DR screening in Poland.
- Authors
Grzybowski, Andrzej; Brona, Piotr
- Abstract
Purpose: We would like to share our experience and results in instigating a pilot diabetic retinopathy (DR) screening programme based on artificial intelligence (AI) in Poland. Diabetes prevalence is increasing and as a result so is DR. Only a few developed countries could so far afford to establish national screening programmes. Those were so far based on work done by ophthalmologists or specialty trained human graders. Methods: We have set‐up our screening programme in a diabetic clinic. Patients who visited the clinic were offered DR screening based on non‐mydriatic retinal images (TRC‐NW400, Topcon). These images were analysed by IDx‐DR, a deep‐learning network based autonomous DR screening system. Based on the IDx‐DR output patients were recommended another screening after 1 year, or an ophthalmology referral with appropriate urgency Results: After an initial period with a number issues related to image quality, staff training, locale etc. we have managed to get the pilot screening programme on working. The AI systems tolerance for image quality was similar to that of a human grader. Compared to an ophthalmologist's grading the IDx‐DR system achived over 90% sensitivity and specificity for referable DR. Conclusions: The cost of a human specialty grader based DR screening programme is prohibitively high for a developing country such as Poland in terms of both human and financial resources. The advent of AI DR screening technology brings a cheaper, easier to roll out alternative that can be successfully employed on a local level. Additionally patients receive a near‐instant result with a satisfactory accuracy. Effectiveness of such system on a larger scale remains to be seen.
- Subjects
POLAND; LARGE scale systems; RETINAL imaging; DIABETIC retinopathy; DEVELOPED countries; ARTIFICIAL intelligence
- Publication
Acta Ophthalmologica (1755375X), 2019, Vol 97, pN.PAG
- ISSN
1755-375X
- Publication type
Article
- DOI
10.1111/j.1755-3768.2019.5078