Training method of structure segmentation model, structure segmentation method and related device

The invention discloses a training method of a structure segmentation model, a structure segmentation method and a related device. Obtaining a plurality of sample image sets, wherein each sample image comprises a labeled structure and an unlabeled structure; inputting the sample image into an initial segmentation network to output a predicted segmentation image of the sample image; and slicing the predicted segmented image into predicted segmented sub-images with labeled structures and predicted segmented sub-images with unlabeled structures. And constructing a prior segmentation image based on the plurality of labeled segmentation images corresponding to the plurality of sample images, inputting the prediction segmentation sub-image of the unlabeled structure and a prior segmentation sub-image of the unlabeled structure in the prior segmentation image into a preset refining network, and outputting a refined segmentation sub-image of the unlabeled structure. And training the initial segmentation network to ob