Transforming Ocular Health: A Vision Transformer Based Eye Disease Detection with GradCam Visualization

Ocular diseases remain a substantial public health concern, necessitating the development of precise and efficient diagnostic methodologies. Notably, despite advances in medical imaging and artificial intelligence, a critical research gap persists in achieving accurate and dependable detection of major eye conditions. In response to this pressing need, our research presents an innovative approach for the identification of five major eye diseases, employing a Vision Transformer (ViT) model in tandem with the GradCam visualization technique. Our method not only bridges this existing research gap but also demonstrates remarkable performance, surpassing the capabilities of existing systems. With an exceptional accuracy of 90.3%, a precision rate of 90.0%, a recall rate of 90.0%, and an F1-score of 90.0%, our proposed method sets a new benchmark for the accurate and reliable diagnosis of ocular conditions. To ensure the robustness and consistency of our approach, we conducted rigorous 5-fold cross-validation, achieving an average accuracy of 89.64%. This research represents a significant advancement in the field of eye disease detection, promising to yield improved clinical outcomes and patient care.