This project presents a fully automatic method for 3D segmentation of brain tissue on MRI scans using a modern deep learning approach and proposes 3D Dense-U-Net neural network architecture using densely connected layers. This project aims to distinguish gliomas which are the most difficult brain tumors to be detected with deep learning algorithms. For a skilled radiologist, analysis of multimodal MRI scans can take up to 20 minutes. Model was trained using the Keras library and final accuracy came out to be 99.82%.