Multi domain feature fusion for Classification of microcalcification In digital mammograms

Abstract

To study the human breast Mammography is used as a diagnostic newlineand a screening tool that uses X rays The objective of mammography is the newlinepremature revealing of breast cancer usually through detection of newlinecharacteristic masses and or microcalcifications Mammography is believed to newlinedecrease mortality from breast cancer This research work deals with newlineclassification of microcalcifications in mammogram based on Jacobi newlinemoments Discrete Wavelet Transform DWT and Singular Value newlineDecomposition SVD newlineMammography is the only effective and viable technique to detect newlinebreast cancer in particular in the case of minimal tumors About 30 to 50 newlineof breast cancers reveal deposits of calcium called micro calcifications A new newlineapproach for detecting microcalcifications in mammograms employing Jacobi newlineMoments is proposed In many applications moments and functions of newlinemoments have been utilized as pattern features Among the various newlineorthogonal moments the Jacobi moments is selected as feature extractor due newlineto its robustness to image noise The set of Jacobi polynomials are orthogonal newlineand this ensures minimal information redundancy between the moments newlineJacobi moments encompass the properties of well known Zernike Legendre newlineand Tchebichef moments These features are used to classify the mammogram newlineimages between normal and abnormal as well as benign and malignant by newlinecomputing the Euclidean distance between the feature vectors newline