AMDNet: Age-related Macular Degeneration diagnosis through retinal Fundus Images using Lightweight Convolutional Neural Network

Age-related Macular Degeneration (AMD) is a retina macular degenerative disease that affects elderly persons. Diagnoses of AMD can be accomplished via manual inspection of typical fundus images. But physicians are limited in their ability to process the full extent of data fundus images provide and their diagnoses are subject to differences in interpretation. This paper proposes an image processing algorithm using a lightweight convolution neural network to improve speed and standardization in AMD diagnosis. The first step in lightweight CNN is a feature extraction algorithm that automatically processes a fundus image to extract important retinal features. In the second step, the proposed method classifies the AMD based on the features extracted in the first step. The proposed network has been trained and tested with STARE and RFMiD fundus databases available publicly. The proposed network has obtained 97.39% and 98.97% accuracy with STARE and RFMiD databases, respectively. The results indicate that the proposed model is lightweight and is better than other state-of-the-art techniques, taken for considerations.