The purpose of our research is to develop non-invasive, computerized methods, based on ultrasound image information, for the computer-aided and automatic diagnosis of some severe affections, such as the diffuse liver diseases (steatosis, hepatitis and cirrhosis), the hepatocellular carcinoma (HCC), the Inflammatory Bowel Diseases and the colorectal tumours. We previously defined (in the context of national research projects) the textural model of these affections, consisting of: the set of the relevant textural features for characterizing these affections; the specific values of the relevant textural features (arithmetic mean, standard deviation, probability distribution). We validated our methods through powerful supervised classification techniques. In this project, we developed advanced, original texture analysis techniques, based on generalized co-occurrence matrices of superior order, in order to improve the accuracy of characterization and classification for these affections. We defined textural microstructure co-occurrence matrices, based on Laws’ and gradient features, and we studied the role of the corresponding Haralick parameters in the supervised classification of the considered affections, as well as in the unsupervised detection of the cirrhosis and HCC evolution phases. Our research, developed at the Computer Science Department, of the Technical University of Cluj-Napoca, was performed in collaboration with the medical specialists from the Medical Imaging Department of the I. Hatieganu University of Medicine and Pharmacy, Cluj-Napoca.