Age-related Macular Degeneration Detection through Fundus Image Analysis Using Image Processing Techniques

Age-Related Macular Degeneration (AMD) is a leading retinal disease that causes vision loss affect people from age fifty five(55) and older. The disease is characterized by the formation of drusen or the yellow deposits containing lipids forming within the macula region of the eye. One of the various ways to diagnose AMD is through obtaining fundus photography using a specialized retinal camera. This study assesses the accuracy of the proposed methodology in recognizing AMD-positive fundus images using Digital Image Processing and various Machine Learning models such as Naïve Bayes (NB), Neural Network (NN), Support Vector Machine (SVM) and Random Forest (RF). The fundus images undergo intensity adjustment and bilateral filter it is then followed by optic disc extraction and Superpixel segmentation using Simple Linear Iterative Clustering. Features, such as Intensity-based statistics and Texton-map Histogram, are extracted and normalized. The resulting values are classified by various Machine Learning algorithms as positive or negative for AMD. The proposed methodology is able to determine Healthy and AMD-positive images while also providing accuracy comparison among Machine Learning models.