Using artificial neural networks for early diagnosis of glaucoma

Purpose : to summarize the experience of development and application artificial neural networks (ANW) in early diagnosis primary open-angle glaucoma (POAG). Material methods . A total 690 patients (918 eyes) were tested. The training clinical group consisted 459 examples (459 eyes), which 369 eyes had an initial stage POAG 90 no glaucoma. testing was represented by 131 (131 110 belonged with 21 without final diagnostic using ANW conducted on 328 unknown researchers, people suspected POAG. complex included optimally necessary set research techniques. Results identified 198 out those (60.4 %) 100 % certainty. 76 (23.2 classified as non-glaucoma, or “healthy”; 54 “doubtful”, whereupon they retested a network pool consisting 5 networks. According results retesting, 28 eyes, 51.9 “doubtful” ones having glaucoma, whereas 26 (48.1 non-glaucomatous, i. e. healthy. Conclusion Our suggests that pose danger doctor patient can be viewed very convenient tool for diagnostics.