Cataract Classification by Applying MobileNetV2 on Different Edge Detection Filters

Advancements in image processing and deep learning have significantly impacted the diagnosis of ocular diseases. Specially, the utilization of edge detection filters is crucial for meticulously segmenting and analyzing vital retinal structures, facilitating early detection and accurate diagnosis of eye diseases. This study investigates the efficacy of edge detection filters in enhancing the quality of fundus images for cataract detection. To do so, we applied the pre-trained deep learning model MobileNetV2. The experiment has been conducted on a real dataset of fundus photographs which is collected from the Combined Military Hospital (CMH), Dhaka. To evaluate the impact of edge detection algorithms, diverse filters have been employed on the image dataset, and their corresponding Mean Squared Error (MSE) and Peak Signal-to-Noise Ratio (PSNR) have been calculated. Our results indicated that the Prewitt filter notably improved the image quality, which, in turn, increased the accuracy of the deep learning model in detecting cataracts.