3D MRI Segmentation using U-Net Architecture for the detection of Brain Tumor
Smarta Sangui, Tamim Iqbal, Piyush Chandra Chandra, Swarup Kr Ghosh, Anupam Ghosh
Abstract
Segmentation of brain tumor from 3D images is one of the most important and difficult tasks in the field of medical image processing as a manual human-assisted categorization can result in incorrect prediction and diagnosis. Furthermore, it is a difficult process when there is a huge amount of data to assist. Extracting brain tumour regions from MRI images becomes challenging due to the great variety of appearances of brain tumours and how similar they are to normal tissues. In this paper, we have designed modified U-Net architecture under a deep-learning framework for the detection and segmentation of brain tumors from MRI images. The applied model has been evaluated on genuine images provided by Medical Image Computing and Computer-Assisted Interventions BRATS 2020 datasets. Test accuracy of 99.4% has been achieved using the above-mentioned dataset. A comparative review with other papers shows our model using U-Net performs better than other deep learning-based models.