Statistical block-based DWT features for digital mammograms classification

Breast cancer is one of the most dangerous types of cancer among women all over the world. If breast cancer is detected in early stage, chances of survival are very high. Mammography is broadly recognized as the most effective imaging modality for early detection of breast abnormalities. Several research works have tried to develop Computer Aided detection/Diagnosis systems (CAD) in order to help radiologists to reduce the variability in the analysis and improve the precision in mammograms interpretation. This paper presents an efficient classification of mammograms using feature extraction. In this approach we propose to use comprehensive statistical Block-Based features, derived from all sub-bands of Discrete Wavelet decomposition. The classification of these features is performed using the Support Vector Machine (SVM). The evaluation of the proposed method is applied on Digital Database For Screening Mammography (DDSM). The system classifies normal from abnormal cases with high accuracy rate (96%). Comparative experiments have been conducted to evaluate our proposed method.