Naslov
Automatic Detection of Microaneurysms in Fundus Images Using an Ensemble-Based Segmentation Method

Autori
Raudonis, Vidas ; Kairys, Arturas ; Verkauskiene, Rasa ; Sokolovska, Jelizaveta ; Petrovski, Goran ; Balciuniene, Vilma Jurate ; Volke, Vallo

Izvornik
Sensors (1424-8220) 23 (2023), 7; 3431, 14

Vrsta, podvrsta i kategorija rada
Radovi u časopisima, članak, znanstveni

Ključne riječi
diabetic retinopathy (DR) ; image segmentation ; microaneurysms (MAs) ; encoder-decoder deep neural network

Sažetak
In this study, a novel method for automatic microaneurysm detection in color fundus images is presented. The proposed method is based on three main steps: (1) image breakdown to smaller image patches, (2) inference to segmentation models, and (3) reconstruction of the predicted segmentation map from output patches. The proposed segmentation method is based on an ensemble of three individual deep networks, such as U-Net, ResNet34-UNet and UNet++. The performance evaluation is based on the calculation of the Dice score and IoU values. The ensemble-based model achieved higher Dice score (0.95) and IoU (0.91) values compared to other network architectures. The proposed ensemble-based model demonstrates the high practical application potential for detection of early-stage diabetic retinopathy in color fundus images.

Izvorni jezik
Engleski

Znanstvena područja
Kliničke medicinske znanosti

POVEZANOST RADA