A deep learning approach for classification of liver disease

Abstract

Liver diseases pose a significant global health burden, with diverse manifestations such as ballooning, fibrosis, inflammation, and steatosis. Accurate and timely diagnosis is crucial for effective treatment planning and patient management. This thesis explores the application of deep learning models, including EfficientNetB2, VGG16, InceptionNetV3, DenseNet121, and ResNet50, for the comprehensive classification of liver diseases based on these distinct pathological features. The study involves a robust dataset of liver pathology images, capturing various stages and manifestations of liver diseases. Through an exhaustive analysis, we compare the performance of different deep learning architectures in accurately identifying and classifying ballooning, fibrosis, inflammation, and steatosis. Our experiments reveal that EfficientNetB2 outperforms the other models in terms of accuracy, demonstrating its efficacy in handling the complexities of liver disease classification. In addition to model performance, the thesis delves into interpretability, providing insights into the features and patterns learned by each model. This contributes to a better understanding of the decision-making process and enhances the clinical relevance of the deep learning models in real-world scenarios. The findings of this research not only showcase the potential of deep learning in liver disease diagnosis but also highlight the significance of selecting appropriate architectures for optimal results. The implementation of EfficientNetB2 in this context opens avenues for improved diagnostic tools and automated systems that can aid healthcare professionals in making more informed decisionsfor patients with liver diseases. The implications of thisstudy extend beyond liver disease classification, emphasizing the broader applicability of deep learning in medical imaging and pathology. The insights gained from this research contribute to the ongoing efforts to enhance the accuracy and efficiency of computer-aided diagnostic systems in the field of hepatology.