A Neural Network Method for Normal/Abnormal Classification of Digitized Mammograms

M.G. Mini and T. Thomas (India)


Automatic detection of breast cancer, Normal mammogram analysis, Probabilistic Neural Network, SGLD matrix.


A novel approach to the problem of computer-aided classification of digitized mammograms for breast cancer detection is presented. The algorithm developed here classifies mammograms into normal and abnormal, based on texture features. The mammograms are pre-processed by identifying and removing normal linear markings formed by the normal connective tissues and using the Spatial Gray-Level Dependence (SGLD) matrix, the texture features are evaluated. These are applied to a Probabilistic Neural Network (PNN) for classification. Using the mammographic data from the Mammographic Image Analysis Society (MIAS) database a true positive recognition score of 94.2% for the abnormal case is achieved.

Important Links:

Go Back