Liver fibrosis detection and classification for shear wave elastography (SWE) images based on convolutional neural network (CNN)

Chronic liver disease (CLD) is a general term that refers to a number of different hepatic disease processes that consume a significant amount of healthcare and financial resources worldwide. Cirrhosis develops when CLD progresses to permanent end-stage liver fibrosis. For the course and prognosis of liver fibrosis, early detection and intervention are crucial. Ultrasound provides a variety of advantages as one of the most commonly utilized methods for detecting liver fibers, including ease, accuracy, and resilience. However, getting subjective and consistent diagnoses is challenging since ultrasound images are inevitably influenced by device characteristics, ultrasound interactions with body tissues, operation procedures, and other uncontrollable elements. The goal of paper is to develop a new proposed liver fibrosis detection model for ultrasound shear wave elastography (SWE) images using deep learning algorithm to distinguish between normal and fibrosis tissue. This paper provides a convolutional neural network-based deep learning feature extraction algorithm for binary classification and detection of liver fibrosis. The dataset used to develop these models was locally collected from an Iraqi hospital. The results obtained revealed that the final model (CNN) was able to classify normal and abnormal liver tissue from SWE with an accuracy of 97.45%, Sensitivity 98.27%, Specificity 96.66%, Precision 96.61% and F1-score of 97.17%. This is accurate results for assist diagnosis liver fibrosis.