Microcalcification diagnosis in digital mammograms based on wavelet analysis and neural networks

The classification of tumors is a medical application that poses a big challenge for in the field of breast cancer detection. The use of artificial intelligence and learning machine techniques has transformed the processes of diagnosis and analysis of breast cancer. For instance, the presence of clustered microcalcifications on X-ray mammograms is vital sign for early detection of breast cancer, which in turn prevents the dire option of surgical breast removal. Digital mammography offers several advantages such as the provision of digital information, which is available in a format that is usable by computer aided diagnosis systems. This work is based on two major approaches; wavelet decomposition analysis and neural network approaches. The system is classified normal from abnormal, mass from microcalcification. Experiments performed on the standard and publicly attainable dataset which is Mammography Image Analysis Society (MIAS), and a comparative analysis is carried out between the test results of this study and recent achievements reviewed in the literature..