Deep coagulation population P system and method for brain metastatic tumor mixed supervised learning

Abstract

The invention provides a deep coagulation population P system and method for brain metastatic tumor mixed supervised learning, and belongs to the technical fields of medical image processing and disease diagnosis. According to the deep coagulation population P system, three modules (including a multi-scale coagulation learning module, a clinical guidance processing module and a refining module) with novel rules and membrane system structures are established, and are realized on a membrane by adopting an end-to-end method, wherein the multi-scale coacervation learning module can perform accurate clustering according to sizes of brain metastasis and perform regression according to a position of a brain metastasis tumor and segmented pixels; and the clinical guidance processing module and the refining module make full use of experience of radiologists and context information of adjacent slices to improve the segmentation accuracy, so that more accurate and efficient brain metastatic tumor segmentation is finally achieved. Therefore, the system has potential clinical application values.

Description

technical field [0001] The invention belongs to the technical field of medical image processing and disease diagnosis, and in particular relates to a system and method for deeply cohesive population P for mixed supervised learning of brain metastases. Background technique [0002] The information disclosed in this background section is only intended to increase the understanding of the general background of the present invention, and is not necessarily taken as an acknowledgment or any form of suggestion that the information constitutes the prior art already known to those skilled in the art. [0003] Automatic detection and segmentation of brain metastases (BM) is crucial for early diagnosis and treatment of BM and helps to improve patient survival. However, the large variability in the size, location, and shape of brain metastases, as well as the low contrast between brain metastases and their surroundings, especially peritumoral edema, make this task challenging. [0004...

AI Technical Summary

Problems solved by technology

However, the inventors found that the above model directly segmented the brain metastases, lacking the agglomerative learning of the brain metastases, while the non-brain metastases regions occupied larger regions in MRI would destroy the agglomerative learning of the brain metastases

Therefore, segmentation results are often unsatisfactory

In addition, different initialization models lead to different segmentation performance, and ignore important domain and expert knowledge related to the specific task of medical diagnosis

Images