ISSN: 2165- 7866
Probuddha Konwar, Julius Bhadra, Manash Jyoti Dutta, Jintu Dowari
In brain tumors treatment planning and quantitative evaluation, determining the tumors extent is a major challenge. Noninvasive Magnetic Resonance Imaging (MRI) has developed as a front-line diagnostic technique for brain malignancies without the use of ionizing radiation. Segmenting the extent of a brain tumor manually from 3D MRI volumes is a time-consuming process that relies greatly on operator competence. For correct tumors extent evaluation, a reliable fully automated brain tumors segmentation approach is required in this scenario. We present a completely automated brain tumors segmentation method based on U-Net deep convolutional networks in this paper. The Multimodal Brain Tumor Image Segmentation (BRATS 2015) datasets were utilized to test our approach, which included 220 high-grade brain tumors.