RGA-Unet: An improved U-net segmentation model based on residual grouped convolution and convolutional block attention module for brain tumor MRI image segmentation

In recent years, brain tumor diseases have seriously threatened brain health. With the rapid development of medical imaging diagnostic technology, magnetic resonance imaging (MRI) has become the preferred imaging method for the diagnosis and treatment of brain tumor diseases. However, the target structure of brain tumors is complex and individual variation is great, which makes the detection and treatment of brain tumors very challenging. To solve the problem of brain tumor MRI image segmentation, we propose a U-net segmentation network based on residual grouped convolution and convolutional block attention module (CBAM), named RGA-Unet. Through the experiment on The Cancer Imaging Archive (TCIA) brain tumor image dataset, the Dice score of segmentation results reaches 95.904%, which is 10.761% higher than that of a traditional U-net network. The experimental results show that this method can effectively realize the automatic and accurate segmentation of brain tumor images, which has certain research significance.