Review of Automatic Segmentation of MRI Based Brain Tumour using U-Net Architecture

Abstract

Segmentation of brain tumour from the surrounding healthy brain tissues by a radiologist is a tedious task. For a safe brain surgery, it is essential to define the contour of brain tumour, for complete resection of tumour. Active research is being carried out in automatic tumour segmentation using deep learning networks for precise segmentation of tumour components. Deep learning networks are more effective at learning patterns and the performance of deep network increases when trained with more data. This paper reviews the automated segmentation of brain tumour in MRI images by using U-Net, and the deep learning network architecture. U-Net is a convolutional neural network architecture developed for segmentation of biomedical images.