Computer Aided Detection System using Entropy Based Segmentation as Decision Support for Detection of Neoplasms and Comparison of Feature Spaces for Classification

Objective: To design a computer aided detection system for the early of breast cancer from mammograms as it can assist doctors in diagnosis. Methodology: The proposed method used Computer-Aided Detection (CAD) is based on using textural differences abnormal and normal to detect cancer. Gabor Features gray-level co-occurrence matrix (GLCM) features are extracted region interest segmented Entropy segmentation. Support Vector Machine (SVM) classifier classifying into cancerous non-cancerous cases. 35 number taken Mammographic Image Analysis Society (MIAS) data set. MIAS database chosen carries more challenging because lot unwanted tissues flesh part included which has intensity level than micro-calcifications. Findings: classification accuracies obtained 92.98% 98.11% respectively. sensitivity achieved with 100% shows no missed case. accuracy higher compared superiority these capturing texture mammograms. Novelty: removal pectoral muscle an important pre-processing step. Only proper elimination muscle, segmentation possible. remove this paper removed muscles all study. entropy technique non-contributing outperforms other CAD systems literature available. These very promising results successful Computer-aided that be put clinical trials or double reading. Keywords: Breast Cancer; Computer Aided System; Segmentation; Features; GrayLevel CoOccurrence Matrix