Learning Contextual and Attentive Information for Brain Tumor Segmentation

Thanks to the powerful representation learning ability, convolutional neural network has been an effective tool for the brain tumor segmentation task. In this work, we design multiple deep architectures of varied structures to learning contextual and attentive information, then ensemble the predictions of these models to obtain more robust segmentation results. In this way, the risk of overfitting in segmentation is reduced. Experimental results on validation dataset of BraTS 2018 challenge demonstrate that the proposed method can achieve good performance with average Dice scores of 0.8136, 0.9095 and 0.8651 for enhancing tumor, whole tumor and tumor core, respectively. The corresponding scores for BraTS 2018 testing set are 0.7775, 0.8842 and 0.7960, respectively, winning the third position in the BraTS 2018 competition among 64 participating teams.