Breast Cancer Diagnosis in Digital Mammogram using Statistical Features and Neural Network

摘要

In this study, the mammogram is classified as either normal or cancer pattern. In the last few decades soft computing improves the accuracy of the breast cancer detection in digital mammograms. The standard approach for diagnosis of breast cancer is biopsy. But biopsy makes patient discomfort, bleeding and infection. The CAD (Computer Aided Diagnosis) is developed for the reason of avoid unnecessary biopsy. The statistical features are extracted from the digital mammograms. These features are fed to neural network classifier to classify it into two classes namely normal and cancer. This study describes neural network classification technique. Experiments have been conducted on images of DDSM (Digital Database for Screening Mammography) database. The performance measures are evaluated by confusion matrix. By increasing the training samples this study reveals the improved classification accuracy. This CAD system achieved 94% accuracy, 96% sensitivity and 92% specificity for diagnosis of breast cancer.