Mutual gain adaptive network for segmenting brain stroke lesions

Boyu Huang, Guanru Tan, Haowen Dou, Zhihan Cui, Youyi Song, Teng Zhou

Abstract: Highlights•An effective and generic learning network for brain stroke lesion segmentation.•A mutual gain adaptive similarity (MGAS) block for segmenting stroke lesions with different sizes.•A GCA block for facilitating the restoration of pixel localization to distinguish ambiguous boundaries.•Extensive experiments demonstrate the outperformance compared with state-of-the-art methods.