Microcalcification Detection in Mammography using Wavelet Transform and Statistical Parameters

Abstract

The earliest sign of breast cancer is the existence of microcalcifications which are tiny calcium clusters in breast tissues detected in mammographies. Early detection and diagnosis of microcalcifications is the main step to improve prognosis of breast cancer, which is one of the most frequently serious disease among women. In this work, we study the methodology based on Bi-dimensional discrete wavelet transform and statistical measurements to estimate the position of these tiny clusters in mammographies. The statistical analysis involves calculating skewness and kurtosis values of all three sets of wavelet coefficients. The crossing of rows and columns associated to the high skewness and kurtosis values determine regions of microcalcifications clusters. Simulation results show that the investigated methodology is successful in the majority of the 18 analyzed images containing tumors.