An Intelligent Breast Cancer Forecasting System using Optimized Elman Deep Neural Network

Abstract

Breast cancer is one of the widely occurring cancer variety among the women and the mortality is assured by early detection. Computer Aided Design (CAD) and computational approaches has significance in the breast cancer detection. The mammogram images have much information including those which will not help in precise identification of region of interest. The feature extraction and selection are vital for any classification problems. The performance of the classifier is highly dependent on the optimal feature set selected for the classification. Hence to avoid high dimensionality data, Grey Level Co-Occurrence Matrix (GLCM) is used and feature reduction is done by Singular Value Decomposition (SVD) where the classification is done by optimized Elman Deep Neural Network (OELDNN). The main intent of optimizing the OELDNN is to enhance the accuracy of classification. It has been shown in the results that the classification accuracy of OELDNN classifier performs better by 6.43% than naive bayes, 3.61% than SVM and 3.11% than CNN.