Brain Tumor Segmentation Using Inception Modules

Among brain tumors, gliomas are the most common primary brain malignancies and they are very aggressive, thus leading to a very short life expectancy in their highest grade. Therefore, accurate and robust tumor segmentation is key stage for diagnosis, treatment planning and risk factor identification. Magnetic resonance imaging (MRI) is a widely used imaging technique to assess these tumors, but the large amount of MR images generated in clinical routine makes it difficult for manual segmentation. In addition, manual segmentation is time consuming, subjective and depends on the level of individual’s experience. Therefore, automatic and reliable segmentation methods are required; however, the large spatial and structural variability among brain tumors make automatic segmentation a challenging problem. In this paper, we propose a novel automatic segmentation method based on Convolutional Neural Networks (CNN). Our method is a combination of U-net and Inception modules. Experiments with BraTS 2020 training set, our proposed method achieved average Dice scores of 0.902, 0.797, 0.855 for whole tumor, enhancing tumor core and tumor core respectively. In this work we show that segmentation results can be improved by adding Inception modules to the U-net.