Automatic Liver CT-Scan Image Segmentation for Tumor Area Detection based on Deep Generative Adversarial Networks

The liver is one of the most vital organs of the body and plays an important role in purifying the blood and removing waste products from the body. Liver like other organs of the body can be disturbed by diseases such as hepatitis, fatty liver and cancer which cannot perform its function well. Liver cancer is one of the deadliest types of cancer and liver-related diseases that threatens the lives of thousands of people around the world every year. Early diagnosis of liver cancer can make treatment steps effective and increase the possibility of saving the patient's life. Image processing is known as one of the effective methods of diagnosing various diseases in which patient data is analyzed and their hidden and useful patterns are discovered for disease diagnosis. Deep learning is one of the techniques that have been used in image processing in recent years, which can be used to diagnose various diseases, including liver cancer. This research uses CT scan data from Data Science Bowl ۲۰۱۷ dataset and the proposed method of this research is based on Deep Generative Adversarial Networks (DGAN) which optimized with Watershed algorithm for better segmentation and detect the exact area. The results show the proposed approach is able to evaluate up to ۹۸.۶۴% accuracy for detecting exact areas of tumors in liver CT scan images.

کلیدواژه ها:

Liver Cancer ، Image Segmentation ، Deep Generative Adversarial Networks (DGAN) ، Deep Learning ، Watershed Algorithm