Supervised classification and prediction of fibrosis seriousness using ultrasonic images

Liver fibrosis is the natural wound healing response to parenchymal injury in chronic liver diseases and may eventually result in liver cirrhosis. Noninvasive imaging methods widely used for the diagnosis of liver fibrosis are Magnetic Resonance Imaging (MRI), Computed tomography (CT), Ultrasound and Elastography. This work aims to extract texture features from ultrasonic liver images. The Artificial Neural Networks(ANN), Gaussian Mixture Model (GMM) and Support Vector Machine (SVM) classifiers were used to classify degree of the liver fibrosis seriousness (F0-F1/F2-F4). The output of the three classifiers were compared so as to evaluate the classification, detection and prediction of liver diseases. The experimental results suggest the effectiveness of the proposed method, as it achieves better total accuracy than those of the other methods.