Brain tumor segmentation using CNN

Abstract

The determination of tumor extent is a major challenging task in brain tumor planning and quantitative evaluation. Magnetic Resonance Imag-ing (MRI) is one of the non-invasive technique that has emanated as a front line diagnostic tool for brain tumor without ionizing radiation. Among brain tumors, gliomas are the most common aggressive, leading to a very short life expectancy in their highest grade. In the clinical prac-tice manual segmentation is a time consuming task and their perfor-mance is highly depended on the operator’s experience. This project proposes fully automatic segmentation of brain tumor using convolutional neural network. Further, it uses high grade gliomas brain image from BRATS 2015 database. The project accomplishes brain tu-mor segmentation using tensor flow, in which the anaconda frameworks are used to implement high level mathematical functions. The survival rates of patients are improved by early diagnosis of brain tumor. Hence, this project segments brain tumor into four classes like edema, non-enhancing tumor, enhancing tumor and necrotic tumor. Brain tu-mor segmentation needs to separate healthy tissues from tumor regions such as advancing tumor, necrotic core and surrounding edema. This is an essential step in diagnosis and treatment planning, both of which need to take place quickly in case of a malignancy in order to maximize the likelihood of successful treatment.