MedFPNet: A Medical Image Segmentation Network Based on Fourier Transform
摘要
Semantic image segmentation is the process of labeling each pixel of an image with its corresponding class. U-Net and its variants are currently the mainstream strategies to segment the medical images. In order to improve the performance of convolutional neural networks on various segmentation tasks, we propose a two-branch segmentation strategy from coarse to fine. MedFPNet, a new architecture based on the Fourier transform, is proposed in which the residual amplitude and the phase blocks are constructed. Experimental results show that our model is superior to U-Net and some of its variants, and it has also gained better performance than some transformer-based models.