Artificial intelligence: Assisted fundus image analysis for medical diagnostics in conflict zones
Artykuł w czasopiśmie
MNiSW
5
spoza listy
| Status: | |
| Autorzy: | Matysiak Magdalena, Podkowiński Arkadiusz, Chorągiewicz Tomasz, Karpiński Robert, Dolecki Michał, Stęgierski Rafał, Zimenkovskyy Andrii, Shybinskyy Volodymir, Jonak Katarzyna E., Rejdak Robert |
| Dyscypliny: | |
| Aby zobaczyć szczegóły należy się zalogować. | |
| Rok wydania: | 2026 |
| Wersja dokumentu: | Drukowana | Elektroniczna |
| Język: | angielski |
| Numer czasopisma: | 1 |
| Wolumen/Tom: | 20 |
| Strony: | 510 - 524 |
| Bazy: | BazTech |
| Efekt badań statutowych | NIE |
| Finansowanie: | Project financed by the Polish National Agency for Academic Exchange under the Urgency Grant program. Title “Cross-border development telemedicine system on the example of ophthalmological examinations in conditions of armed conflict - analyzes and recommendations.”, project number: BPN_GIN_2022_1_00120. |
| Materiał konferencyjny: | NIE |
| Publikacja OA: | TAK |
| Licencja: | |
| Sposób udostępnienia: | Otwarte czasopismo |
| Wersja tekstu: | Ostateczna wersja opublikowana |
| Czas opublikowania: | W momencie opublikowania |
| Data opublikowania w OA: | 21 listopada 2025 |
| Abstrakty: | angielski |
| Artificial intelligence (AI) has become an important tool for recognizing changes in the ocular fundus, but most existing studies are conducted in peacetime clinical environments with advanced diagnostic equipment and stable infrastructure. In contrast, wartime conditions impose severe constraints, including limited access to sophisticated imaging devices, reduced medical resources, and the urgent need for rapid decision-making. This article addresses this research gap by examining AI-assisted classification of retinal fundus images collected under conflict conditions in Ukraine. Three approaches were employed: feature extraction combined with deep neural networks, convolutional neural network (CNN)-based models, and Microsoft’s Custom Vision platform. The dataset consisted of 448 retinal images divided into five groups: normal findings, trauma-related injuries, optic nerve disc changes, vascular lesions, and macular degeneration. Despite the small and imbalanced dataset, and the challenging acquisition environment, each pre-processing method achieved at least 80% classification accuracy, with the CLAHE method yielding the best results. This study demonstrates, for the first time, that AI can provide reliable ophthalmic diagnostics in extreme and resource-limited wartime settings, bridging the gap between peacetime and conflict healthcare. |
