Computing Texture Features Based on Information Theory from Co-occurrence Matrix in Digitized Mammograms

摘要

In this paper, spatial gray level co-occurrence matrix is used to extract statistical texture features in digital mammograms. Measures of texture computed using only gray level histogram moments statistical texture analysis method suffer from the limitation that they depend only on individual pixel values and not on the interaction or cooccurrence of neighboring pixel values. To solve this problem, spatial gray level co-occurrence matrices are used to extract statistical texture features in digital mammograms. Information theory based features are important texture features, which are computed from spatial gray level co-occurrence matrix, to build a robust descriptor towards correctly classifying abnormal and normal regions of mammograms. Entropy is a most important statistical texture feature based on information theory. Entropy measures the randomness of intensity distribution. In most feature descriptors, Shannon's measure is used to measure entropy. In this paper, Havrda and Charvat 's measure is also used to measure entropy. Havrda and Charvat's entropy has a higher dynamic range than Shannon entropy over a range of scattering conditions, and are therefore useful in estimating scatter density and regularity. Experiments have been conducted on images of mini-MIAS database (Mammogram Image Analysis Society database (UK)). The Results of this study are quite promising. This work is a part of developing a computer aided decision (CAD) system for early detection of breast cancer.