Feature Extraction and Classification for Human Brain CT Images

In this paper, computer aided diagnosis (CAD) is applied to the brain CT image processing. In addition to the 3 classical types of features, i.e. gray scale, shape and texture, the symmetric feature based on the characteristics of human-brain CT image is extracted. Inductive learning techniques, See5 and RBFNN (radial basis function of nerve network) are used to build classifiers for normal and abnormal brain CT images. Experimental results show that CAD system of human-brain CT image processing can assist doctors to correctly classify the CT images.