Method and device for automatic extraction of blood vessel centerline

Abstract

An embodiment of the present invention provides a method and device for automatically extracting a blood vessel centerline, comprising: acquiring an original image including a blood vessel structure, and performing multi-scale and multi-directional filtering on the original image based on convex symmetry to obtain an enhanced centerline image; Based on the center line enhanced image, through the non-maximum value suppression method, a set of seed points whose confidence meets the preset condition is obtained, and a set of high-confidence seed points is constructed; through preset rules, the set of high-confidence seed points is sequentially selected. Based on the seed points in the centerline enhanced image, the minimum path method based on backtracking path propagation is used to search for the multi-branch centerline structure of the tubular structure, and the initial centerline structure of the original image is obtained, and then the minimum path method based on key point detection is used to detect all The initial centerline structure is repaired to obtain the complete centerline of the original image. It avoids the large demand for seed points in multi-branch structure extraction based on the minimum path method.

Description

technical field [0001] Embodiments of the present invention relate to the field of image processing, and in particular to a method and device for automatically extracting a blood vessel centerline. Background technique [0002] Tubular structures appear in images in various fields, such as coronary vessels in medical images, retinal vessels, road structures in aerial images, etc. In medical images, the analysis of tubular structures is a necessary link in understanding vascular lesions and describing vascular diseases, among which, the extraction of the centerline of tubular structures is a very important research content. The centerline can represent the topology of the vascular network, and its completeness and accuracy affect the validity and robustness of various applications. [0003] Since the centerline of the tubular structure can completely express the topology of the tubular structure, the automatic extraction technology of the vascular centerline has been widely ...

AI Technical Summary

Problems solved by technology

However, tubular structures usually have complex topological structures and large changes in radius scales, and their imaging images often have problems such as noise, uneven gray scale, complex background, low contrast and low resolution, which lead to difficulties in accurately extracting the centerline of tubular structures. huge challenge

Images