Texture Classification Using Fuzzy Cognitive Maps for Grading Breast Tumor
R. Roopa Chandrika,,
Medical decision support system is a complex medical image analysis system that requires an efficient pattern classification tool that is easier to represent and to perform better classification of abnormalities present in medical images. Fuzzy Cognitive Map (FCM) is a simple, efficient cognitive tool used recently to model such complex and dynamic systems. FCM is integrated with medical decision support system that requires grading of suspicious tissues present in human body. FCM is used in this work to grade suspicious breast cancer cells with the texture properties extracted from digital mammograms. The map is constructed using the texture properties as its concepts and are interconnected based on the causal relationship among the concepts. The patterns or the features extracted from the digital mammogram are based on statistical measures suitable to distinguish between normal and abnormal tissues. GLCM (Gray Level Co-occurrence Matrix) and Laws energy measures are statistical methods used in this work to obtain the textural features. The texture concepts used as input for the FCM tool have shown to classify the severity of abnormality present in digital mammograms better than the other classifiers that used training algorithms like neural network, decision trees etc. The outcome of the automated reasoning of FCM is similar to the qualitative assessment tool used by the medical experts.