An efficient CAD system for detection and classification of tumors in mammographic images using variety features and Probabilistic Neural Network

This research aims at segmentation of the Breast Masses from Digital Mammograms and their classification using Probabilistic Neural Network. The Mammograms of different patients with Fibroadenoma and M Invasive Ductal Carcinoma type of tumor are considered. The work proposed consists of different stages, namely, preprocessing, segmentation, feature extraction and classification. Segmentation of the tumors from the digital mammograms is done using three methods, namely, Local Thresholding, Mathematical Morphology and LBG algorithm. Different statistical, textural and shape features are extracted from the segmented tumor. The varieties of features are extracted from the known tumors and these features are used to train the Probabilistic Neural Network. The system efficiently classifies the tumor into Fibroadenoma, a benign type of breast tumor, and M Invasive Ductal Carcinoma which is a malignant breast tumor.