Ultrasound-Based Hepatitis C Stage Classification Using Deep Learning Techniques

Abstract

Hepatitis C virus (HCV) infections represent a significant threat to public health worldwide, highlighting the need for precise and effective diagnostic techniques for disease staging. This research presents a deep learning methodology utilizing This work employs Convolutional Neural Networks (CNNs) to automatically determine the stages of Hepatitis C Virus (HCV) based on ultrasound imaging. It integrates state-of-the-art architectures such as Xception, DenseNet, ShuffleNet, and ManualNet to improve diagnostic precision and streamline classification processes. The dataset includes a wide range of ultrasound images that highlight essential biomarkers related to disease progression. Among the tested models, Xception recorded the highest accuracy (99.12%), followed closely by DenseNet (98.76%), ShuffleNet (96.89%), and ManualNet (94.32%). The proposed models outperformed traditional machine learning algorithms, achieving higher accuracy than Support Vector Classifiers (92.34%) and Logistic Regression (89.45%). Their robustness was further confirmed through evaluation metrics such as the Area Under the ROC Curve (AUC), Mean Absolute Error (MAE), and Mean Squared Error (MSE), highlighting the effectiveness and reliability of the deep learning approaches.This AI-driven system facilitates early detection of liver damage, aids timely medical intervention, and enhances precision medicine by uncovering imaging patterns often missed by radiologists. Its integration into clinical workflows significantly improves patient care, optimizes prognosis, and supports personalized treatment strategies, revolutionizing Hepatitis C diagnostics.