Evaluation of liver fibrosis based on ultrasound radio frequency signals

Liver fibrosis is a common consequence of almost all chronic liver diseases. Precise and timely evaluation of liver fibrosis progression is essential for the treatment of liver disease. In this paper, we present a deep learning-based framework for evaluating the degree of liver fibrosis, using ultrasound radio frequency signals. The deep learning model based on long short-term memory (LSTM) network and attention cell. The dataset consisted of 96 sets of ultrasound radio frequency signals of rat livers, with five fibrosis stages ranging from 0 to 4. The accuracy and the areas under the receiver operating characteristic curve of the four classification models were greater than 0.81 and 0.89, respectively. This study indicated that evaluation system of liver fibrosis stage based on deep learning approaches and ultrasound RF signals was promising and it would be of great value in monitoring liver fibrosis precisely and non-invasively.