CFFormer: Cross CNN-Transformer channel attention and spatial feature fusion for improved segmentation of heterogeneous medical images

Jiaxuan Li, Qing Xu, Xiangjian He, Ziyu Liu, Daokun Zhang, Ruili Wang, Rong Qu, Guoping Qiu

Abstract: Highlights•CFFormer achieves outstanding segmentation performance, particularly on medical images with blurry boundaries and low contrast.•The proposed Cross Feature Channel Attention (CFCA) module enhances the model’s ability to selectively exchange information between feature maps across different encoder layers.•The proposed X-Spatial Feature Fusion (XFF) module effectively eliminates semantic discrepancies between encoders while enabling efficient feature integration.•CFFormer outperforms state-of-the-art (SOTA) methods across eight datasets spanning five different imaging modalities.

External IDs:dblp:journals/eswa/LiXHLZWQQ26