Multi-organ CT Image Segmentation Model Based on Swin Transformer and CNN

With the rapid development of medical imaging technology, Computed Tomography (CT) has become an indispensable tool in medical diagnosis. However, in the application of medical image segmentation, there are still some problems in CT images, such as the diversity of organ shape, size, location, texture, and interference factors like noise and artifacts. In response to these problems, this paper proposes a U-shaped medical image segmentation network STC-Unet based on a combination of CNN and Transformer. Based on the traditional U-Net with residual structure, the ST module improved by Swin Transformer was added, carry out global feature extraction and capture and be able to extract multi-scale spatial information, and add the Spatial Pyramid Pooling Coordinate Attention (SPP-CA) attention mechanism to the decoder to increase the module's ability to integrate spatial information. Relevant experiments were conducted on the Synapse multi-organ segmentation dataset. Experimental results show that STC-Unet has a DSC of 83.99% and HD95 of 15.59mm, achieving better segmentation results in multi-organ segmentation.