A Survey of Semantic Segmentation on Biomedical Images Using Deep Learning

Abstract

This survey presents an overview of recent attempts on semantic segmentation on biomedical images using deep learning techniques. Deep learning is a revolutionary technology and recent advances in it have made it possible to improve the performance of semantic segmentation methods, which are common tasks in medical imaging. Recently, semantic segmentation is a breakthrough in biomedical image segmentation. Semantic segmentation is the process of classifying each and every pixel of an image into a class. This paper explicitly deals with various semantic segmentation methods applied to different biomedical images. Further, diverse architectures of deep neural networks for semantic segmentation of images such as U-Net, fully convolution network (FCN), and SegNet are examined. In brief, datasets namely Dristhi-GS, ALL-IDB1, H & E stained images, Flair-MRI Brats 2015, and Japanese Society of Radiological Technology (JSRT) are studied. In this paper, a complete taxonomy of various methods of semantic segmentation for medical images is given.