MEDICAL IMAGE SEGMENTATION USING U-NET

Abstract

U-net is an image segmentation method that was created mainly for image analysis in medicine. It is very useful in the field of medical imaging since it can segment images accurately with just a modest quantity of training data. As a result, U-net is now widely used as the main tool for medical imaging segmentation tasks. In this paper, we explore recent trends and analyze the different developments in the U-net architecture. We also compare technologies based on several variations of the U-net structure and give common evaluation metrics. The performance of the U-Net architecture was validated by its experimental results on two distinct datasets: one for skin cancer segmentation and the other for nuclei segmentation. The accuracy and loss metrics showed satisfactory performance, thus confirming the U-Net's usefulness in the medical field. Finally, we provide a useful reference for future research by outlining possible future study topics and challenges.