H$$^2$$NF-Net for Brain Tumor Segmentation Using Multimodal MR Imaging: 2nd Place Solution to BraTS Challenge 2020 Segmentation Task

In this paper, we propose a Hybrid High-resolution and Non-local Feature Network (H $$^2$$ NF-Net) to segment brain tumor in multimodal MR images. Our H NF-Net uses the single cascaded HNF-Nets different sub-regions combines predictions together as final segmentation. We trained evaluated our model on Multimodal Brain Tumor Segmentation Challenge (BraTS) 2020 dataset. The results test set show that combination of models achieved average Dice scores 0.78751, 0.91290, 0.85461, well Hausdorff distances ( $$95\%$$ ) 26.57525, 4.18426, 4.97162 for enhancing tumor, whole core, respectively. method won second place BraTS challenge segmentation task out nearly 80 participants.