Fast detection of masses in computer-aided mammography

A method based on the radial basis function neural network and a set of decision criteria is presented for detecting circumscribed masses in mammograms. This method is implemented by taking into account the multi-scale statistical properties of the breast tissue, and succeeds in finding the exact tumor position by performing sub-image windowing analysis. It is shown that fast implementation in both feature extraction and the neural classification module can be achieved. The circumscribed mass detector is implemented in the MIAS-database, giving reliable detection accuracy in both position and mass size.