Radial-basis-function based classification of mammographic microcalcifications using texture features

Mammography has been established as the only effective and viable technique to detect breast cancer especially in the case of nonpalpable and minimal tumors. About 30% to 50% of breast cancers demonstrate deposits of calcium called microcalcifications. We investigate the potential of using textural features for their correlation with malignancy. A combination of global texture features extracted from the second histogram was combined with local texture features obtained from a wavelet decomposition of the regions containing the calcifications. The performance of the radial-basis-function neural network was compared to the standard multilayered perceptron. The neural networks yielded good results for the classification of hard-to-diagnose cases of mammographic microcalcification into benign and malignant categories using the selected set of features.