Singular Value Decomposition Based Detection of Diabetic Retinopathy

Author (s):

  1. Amol P. Bhatkar, Anuradha Engineering College, Chikhli, Maharastra, India,
  2. Dr. G. U. Kharat, Sharadchandra Pawar College of Engineering, Otur, India


An automated image analysis method to detect early signs of diabetic retinopathy is demand for today’s world. This paper focuses on Generalized Feed Forward Neural Network (GFFNN) to detect diabetic retinopathy in retinal images. In this paper we used GFFNN classifier to classify the retinal images into normal and abnormal. 64-point Singular Value decomposition (SVD) with 09 different statistical parameters is used to form the SVD based feature vector. This feature vector is used as an input GFFNN classifier. The % classification accuracy is 98.97 and 95.83% on train and CV datasets respectively. The sensitivity and specificity of GFFNN classifier are 91.67% and 100% respectively.

No of downloads: 25