An artificial neural network for detecting microcalcifications in wavelet-enhanced digitised mammograms

The presence of microcalcification clusters (MCCs) is a primary sign of breast cancer. Thus, the successful detection of microcalcifications during mammographic examination is vital for the early diagnosis of the cancer. Computer-based detection methods aim to ameliorate the diagnostic process by providing the radiologist with a second opinion. An automatic detection scheme making use of a neural network classifier, with input feature vectors containing spatial and spectral image attributes, is investigated. A wavelet-based image enhancement technique is employed to improve the detection. The detection scheme is tested and preliminary results are reported.