Training method of image segmentation model, and image segmentation method and device

The invention relates to a training method of an image segmentation model, and an image segmentation method and device. The method comprises the steps of obtaining sample data; the sample data comprises a medical image of a sample object and an initial segmentation region of the sample object; wherein the initial segmentation region comprises an initial outermost layer segmentation region, and the initial segmentation region is used for representing a segmentation region preset for the sample object; generating a pseudo segmentation region according to the initial outermost segmentation region, and taking the pseudo segmentation region as a target outermost segmentation region; and taking the medical image as input, taking the initial outermost layer segmentation region and the target outermost layer segmentation region as supervision information, and training a to-be-trained image segmentation model to obtain a trained image segmentation model. By adopting the method, the problem that the actual segmentation