MEDICAL IMAGE CLASSIFICATION FOR DISEASE DIAGNOSIS BY DBN METHODS

Abstract

A radical deep learning-based feature extraction approach for disease diagnosis is discussed in this paper. This
approach focusses on the development of automatic screening system which is capable of diagnosing diseases such
as collateral disease, retinal disease, drain, heart diseases etc. Some of these diseases shares common characteristics,
which makes their classification difficult. In an effort to subdue the aforementioned problem, DBN (Deep Belief
network) in association with a multi class SVM classifier is utilized. The main contribution of this work is the
reduction of complexity in the process of finding the significant features and thus reducing its dimensions to classify
the nature of the disease. The paper depicts efficient feature extraction methods for diagnosis of retinal diseases.