A Novel Approach for Mass Classification in DigitalMammogram Using Multiresolution Analysis andAdaptive Dimension Reduction

A novel approach for mass classification of breast cancer in digital mammogram based on Discrete Wavelet Transform (DWT) and Adaptive Dimension Reduction (ADR) technique is presented here. The mass classification is obtained by using DWT at various levels for decomposition. Then reduction of the high dimensional wavelet coefficients is done by using ADR. Dimension reduction and unsupervised learning (clustering) are combined in ADR. Classification of the mammogram image into normal, benign or malignant is done using this reduced wavelet coefficients as features. In the proposed system KNN (K-Nearest Neighbor) and SVM (Support Vector Machine) classifiers are used to classify the mammograms.