A Liver Cirrhosis Segmentation and Detection Using Modified Deep Learning Model

Abstract

This study uses machine learning and deep learning, including ResNet50, XGBoost, and Random Forest, to identify liver cirrhosis. Severe liver cirrhosis requires early identification and treatment. Traditional diagnostic methods work but take time and may be unclear. The deep convolutional neural network ResNet50 automatically recognizes complicated medical imaging patterns for accurate diagnosis. We trained the ResNet50 model on a large liver imaging dataset to distinguish between cirrhotic and non-cirrhotic liver tissues. We also used XGBoost and Random Forest classifiers to improve prediction. The ResNet50 model with XGBoost and Random Forest classifiers was more accurate, sensitive, and specific than other diagnostic methods that were already in use. These powerful machine learning and deep learning models might enhance screening and help doctors make rapid, accurate diagnoses. This study demonstrates that ResNet50, XGBoost, and Random Forest may improve liver cirrhosis detection, improving patient outcomes and lowering healthcare expenditures.