AN U-NET BASED EFFICIENT ARCHITECTURE FOR MEDICAL IMAGE SEGMENTATION

Abstract

In today’s generation, where the medical science also believes in digital technology for diagnosis, there is a lot of digital data which needs to be analyzed by the doctors. Biomedical image segmentation finds its vast application in the scenario. It is the process of dividing the image into a multiple set of pixels, thus allowing the doctors to focus on the region of interest. Image segmentation is the basic step used before image classification, thus finds its utmost importance in the medical analysis. As we all know that skin cancer is one of the most toxic and incurable disease, detecting it in the early stage is very essential. In this paper, I would like to carry forward my thesis in the field of skin lesion segmentation as the low contrast between the skin and the infected area, i.e lesion is challenging. In this report, I would like to discuss about the work and the various steps involved in Medical image segmentation. I would like to focus on Further, I present the implementation and results of medical image segmentation. Finally, I conclude by mentioning the observations made and the future scope regarding this work.