CirrhosisX: Detection and Staging of Liver Cirrhosis using DL and Explainable AI

Abstract: Liver cirrhosis is a serious, ongoing health problem that leads to permanent harm in the liver from endless scarring and fibrosis. It hits hard on a global scale, being one of the top reasons people die around the world. Spotting it early and figuring out exactly how far along it is can really change things for patients, cutting down on the chances of more serious problems popping up.

This project introduces a hybrid approach to analyzing the MRI images of patients suffering from liver cirrhosis by using artificial intelligence methods. The DenseNet121 neural network is used to extract features from the medical images of patients, and this extracted information is then used to classify whether the patient is cirrhotic and at what stage this patient is cirrhotic. Not only does this method predict the likelihood of cirrhosis in a patient but also uses the Grad-CAM technique to visualize parts of the MRI images that help make these predictions.The performance of the system proposed herein is evaluated based on traditional performance indicators such as accuracy, precision, recall, and F1-scores.

In conclusion, the suggested approach can be used as an additional instrument that will help clinicians diagnose liver diseases and lessen their work.

Keywords: Liver Cirrhosis, Deep Learning, DenseNet121, XGBoost, Explainable AI, Medical Image Analysis, Grad- CAM, Hybrid Learning Model