Automated Optic Disc Localization from Smartphone-Captured Low Quality Fundus Images using YOLOv8n-Based Model

In areas where conventional fundus cameras are logistically unaffordable, the smartphone-based approach is considered as a promising future of glaucoma screening due to their affordability and ease-of-use. Optic disc localization is a critical stage during automated glaucoma screenings. For fundus images captured with standard fundus cameras, the majority of the models available out there can locate the optic disc with satisfactory performance. However, for images that are captured with smartphone-based fundus cameras, the inherent noise and lower quality of the fundus image makes it difficult for models to accurately locate the optic disc region. In this study, we have proposed to utilize YOLOv8n-based model for optic disc localization due to the model's cutting-edge performance in diverse tasks and its lightweight nature. To further improve the the model's complexity and computational cost, we have replaced the C2f modules at the backbone with convolutional block attention module. In this study, we have used a public dataset which has 2000 low quality fundus images that are captured using a smartphone-based fundoscopy device. From these, 60% of the data was used for fine-tuning the model, and 25% for testing. Using a confidence threshold of 50%, the enhanced model achieved a 100% accuracy rate in detecting the optic disc on the test images. Moreover, over 96 % of these test images attained an intersection over union value exceeding 0.85. The inclusion of the attention model notably reduced the GFLOPs from 8.2 to 2.2, resulting in a less complex model while maintaining nearly similar performance. These results highlight the potential of the lightweight YOLOv8n model for deployment in resource-constrained environments, offering a promising performance on accurately localizing the optic disc and enhancing the feasibility of affordable glaucoma screening on smartphones.