An automated system for classifying computed tomographic liver images

This paper presents an automated system for the classification of different digitized computed tomographic images of the human liver. The proposed system consists of four main steps. First, images were preprocessed to enhance the image contrast and segment the human liver images from background and surrounding organs. Second, five sets of features were extracted using: (1) statistical-based features, (2) intensity-based approach, (3) morphological-based features, (4)frequency domain-based, and (5)wavelet domain- based features. The features were extracted from each computed tomography (CT) image. In the third step, the set of distinct features, checked by feature selection using Principal Component Analysis (PCA). Finally the selected features are applied to an artificial neural network (ANN) based classifier in order to determine the set of features that distinguishes better between the normal/tumor classes. It has been found that the classifier based on discrete wavelet features reaches a correct classification rate of 95%.