Real-Time disease detection and analysis in optical coherence tomography (OCT) images using an improved YOLO architecture Geliştirilmiş YOLO mimarisi kullanılarak optik koherens tomografi (OCT) görüntülerinde gerçek zamanlı hastalık tespiti ve analizi

Creative Commons License

Türkdoğan H., ÖZCAN T., HOROZOĞLU F.

Journal of the Faculty of Engineering and Architecture of Gazi University, cilt.41, sa.2, ss.1441-1452, 2026 (SCI-Expanded, Scopus, TRDizin)

Özet

Retinal diseases are among the major health conditions that can lead to vision loss by affecting the retina. Early detection of these diseases is critical for preventing permanent visual impairment. Currently, Fundus Fluorescein Angiography (FFA) and Optical Coherence Tomography (OCT) imaging techniques are widely used for the diagnosis of retinal diseases. However, the interpretation of these images requires expert ophthalmologists, and the increasing number of patients, together with the shortage of specialists in some regions, makes the diagnostic process more challenging. In this study, the automatic classification of eight different retinal diseases was performed using the Retinal OCT-C8 dataset. Standard YOLO models as well as improved YOLOv5 and YOLOv8 models were employed during the classification process. The proposed models were enhanced by increasing the depth of the network architecture, while K-Means-based anchor box optimization was additionally applied to the YOLOv5 model. Experimental results demonstrated that the improved YOLOv8 model achieved the best performance, with 98.1% accuracy, 97.6% precision, 98.6% recall, and a 98.0% F1-score. To evaluate the generalization capability of the proposed model, additional experiments were conducted on the OCTDL and OCTID datasets, yielding accuracy rates of 88% and 86%, respectively. Furthermore, a software application was developed to enable physicians to analyze retinal images and manage patient information. The results indicate that the proposed approach provides an effective and reliable solution for the automatic diagnosis of retinal diseases.