Texture Feature Classification of Liver Sonography Using Fuzzy Similarity Measures

It is difficult to differentiate diffuse liver diseases, namely cirrhotic and fatty livers from normal one by visual inspection from the ultrasound images. The need for computerized tissue characterization is thus justified to assist quantitatively the sonographer for accurate differentiation. In this paper a novel approach of tissue characterization using pattern recognition techniques is developed. Textural analysis methods based on cooccurrence matrix and gray-level gradient variations were applied to extract quantitative parameters for over 150 cases of three liver pathologies namely cirrhotic, fatty and normal livers. In addition to these textural feature descriptors an attenuation and speckle parameters were computed from the B-mode images. A fuzzy similarity measures as an approximate reasoning technique of matching between an unknown case defined by a feature vector and a family of prototypes were used for the classification steps. Finally we tested different textural methods and we could obtain a good results ranging from 80-95% of sensitivities and specificity for different liver pathologies.