A network model segmentation method and device for a coronary artery

Abstract

The invention discloses a network model segmentation method and device for a coronary artery. The method comprises the following steps of screening a plurality of coronary artery training samples to obtain a diversity training sample set and a singularity training sample set; obtaining the segmentation prediction results of the basic model and the extended model by using the diversity training sample set and the singularity training sample set respectively; fusing the segmentation prediction result of the basic model and the segmentation prediction result of the extended model to generate a fusion optimized coronary artery segmentation result. The invention effectively combines the advantages of the diversity training sample and the singularity training sample, so that the problem that thenetwork model trained only with the diversity training samples lacks detailed feature expression for the special samples, and the network model trained only with the singleness training samples is not suitable for the diversity of actual production environment, is solved, and the usability and robustness of the artificial neural network in the work of coronary artery segmentation are further improved.

Description

technical field [0001] The invention relates to the technical field of medical imaging, in particular to a network model segmentation method and equipment for coronary arteries. Background technique [0002] In the field of modern medical technology, automatic coronary artery reconstruction technology has important clinical value and practical significance for doctors. To perform automatic coronary artery reconstruction, it is first necessary to solve the problem of automatic coronary artery segmentation in the process. Different from traditional coronary artery segmentation methods, due to the many advantages of artificial neural networks, people are increasingly inclined to use artificial neural networks to complete coronary artery segmentation. [0003] However, artificial neural networks based on deep learning neural networks have a high number of training samples, and at the same time require a large number of diverse training samples and single training samples. When ...

AI Technical Summary

Problems solved by technology

[0003] However, artificial neural networks based on deep learning neural networks have high requirements for the number of training samples, and at the same time require a large number of diverse training samples and single training samples

When the network model obtained by only using diverse training samples for training lacks detailed feature expression for special samples, which affects the results, and the network model obtained by only using single training samples for training, there is diversity that is not suitable for the actual production environment The problem

Images