U-shaped medical image segmentation method and equipment for BiFormer structure backbone extraction

The invention relates to a U-shaped medical image segmentation method for BiFormer structure backbone extraction, and the method comprises the following steps: obtaining medical image data, and building a medical image data set; preprocessing the medical image data set; establishing a medical network segmentation model; executing training of a medical network segmentation model; executing an application of the medical network segmentation model; wherein the step of building the medical network segmentation model further comprises the following steps: setting an encoder, a decoder and a jump connection module in the medical network segmentation model; the encoder is composed of an input layer and an encoding layer of a four-level pyramid structure. And each level of coding layer comprises a Patch Merging layer and a BiFormer layer. According to the method, a U-shaped segmentation network taking BiFormer as a trunk feature extraction module is constructed, and an introduced dynamic attention mechanism BRA modul