Fig. 6From: Aggregation-and-Attention Network for brain tumor segmentationThe architecture of the dual-attention headBack to article page