Authors: Rohit.N, Dr.R. Kannadasan
Abstract: In human anatomy, the liver has a special feature called regeneration, and this feature helps the liver to grow back even after a large part of its organ is removed. Like regeneration, it helps maintain bile salts, protein synthesis, and detoxification. Irregular eating habits, sleep, and alcohol consumption increase liver function and cause various diseases such as fatty liver and other liver problems. Hepatocellular carcinoma is one of the liver diseases, and it is caused by the abnormal growth of cells in the liver. In such cases, liver regeneration is possible if the disease is identified at an early stage. To support this early identification, several research works have been carried out using both artificial intelligence and deep learning techniques. Therefore, this paper proposed an automated approach to identify liver cancer at an early stage through a segmentation and classification using deep learning techniques. The early-stage identification becomes possible with the dual stage of segmentation using U-NET and XAI. The U-NET helps to segment the image through its various texture properties of the image. Then, the XAI is used to analyze the individual regions of the image. This special feature of this approach is that it uses descriptive AI for classification, and this helps in identifying critical regions through heat maps and saliency maps. This technique was tested on a computed tomography dataset of liver images and its performance was evaluated in terms of precision, accuracy, recall and F1-score.
