Abstract
Precise segmentation of brain tumor from magnetic resonance images (MRIs) is of vital importance for lesion measurement, surgical planning decision, and clinical treatment evaluation. In the context of rapid development of deep neural networks (DNNs), increasing number of DNN-based brain tumor segmentation methods have been recently reported. However, most methods usually ignore the interaction among multi-modal data and multi-scale context features, resulting in poor segmentation accuracy on challenging cases, e.g., tumors with complex interactions between sub-regions. This paper proposes a novel Adaptive Context Aggregation Network (ACANet) for multi-modal brain tumor segmentation. ACANet follows a parallel encoder-decoder framework. Specifically, to bridge the gaps between different modalities, we propose an adaptive context aggregation module that adaptively fuses multi-modal features from the encoders of two input branches to explore their complementary information and aggregates multi-scale context information extracted by a set of dilated convolutions. To cope with the complex boundary interactions, we propose a prediction-aware decoding strategy at the decoder to help the network focus more on error-prone regions through the guidance of two temporal predictions when incorporating the features from the encoder. Extensive experiments on two brain tumor segmentation datasets demonstrate the superiority of ACANet over eight mainstream methods. Additionally, a series of ablation studies supports the effectiveness of ACANet’s underlying concepts.
Funding Information
  • National Natural Science Foundation of China (62371305)
  • Natural Science Foundation of Shenzhen (JCYJ20230808105906013)
  • Basic and Applied Basic Research Foundation of Guangdong Province (2024A1515030025, 2024A1515030278, 2022A1515010160)
  • Medicine Plus Program of Shenzhen University (2024YG007)
  • Tencent “Rhinoceros Birds”—Scientific Research Foundation for Young Teachers of Shenzhen University