Information processing method, device and equipment for performing lesion medical image segmentation and medium

Abstract

This application discloses an information processing method, apparatus, device, and medium for lesion medical image segmentation, relating to the field of computer technology and applied to computer devices. The method includes: determining anatomical path prompts corresponding to the medical image to be segmented based on a pre-constructed target medical knowledge graph; the anatomical path is a spatial path reflecting the anatomical relationships between regions, organs, and lesions, and the anatomical path prompts include target spatial position constraints and spatial shape constraints of organs within the path; determining organ pseudomask generation results based on a multimodal medical basic model and the anatomical path prompts; and determining the image segmentation result using a preset segmentation model, a preset evidence constraint processing framework, and the organ pseudomask generation results; the preset evidence constraint processing framework is used to convert segmentation evidence data into constraints. This application improves the accuracy, reliability, interpretability, and clinical applicability of information processing for lesion medical image segmentation.

Images