Blood vessel diameter analysis system based on angiography image

Abstract

The invention discloses a blood vessel diameter analysis system based on an angiography image, and belongs to the field of computer vision and machine learning. The system is composed of four modules. The module 1 is used for automatically segmenting a blood vessel region by using an angiography image; the module 2 is used for carrying out distance transformation analysis according to the blood vessel segmentation result to obtain a distance image from the pixels in the blood vessel to the blood vessel boundary; the module 3 is used for clicking a starting point and an ending point of a local blood vessel section, and designing a path detection optimization target to estimate a motion trail of an inscribed ball passing through a bleeding tube; and the module 4 is used for acquiring a diameter curve of the moving inscribed ball as a blood vessel diameter curve. The invention has the advantages that the explanatory property is high, the calculation is simple, the problem of multi-step error accumulation of a traditional 2D image reconstruction strategy is avoided, the unreasonable lumen morphological assumption of a traditional analysis method is also avoided, and the invention is more suitable for automatic analysis of the diameter of the blood vessel.

Description

technical field [0001] The present invention relates to the fields of computer vision and machine learning, and more specifically, to an automatic analysis system for vessel diameters based on angiography images. Background technique [0002] Arteriographic imaging is a method of imaging blood vessels that highlights arterial vessels. Analyzing the caliber curve of blood vessels based on angiographic images can intuitively and visually display the change rule of vascular caliber, which is an image information visualization technology. [0003] The mainstream vascular diameter analysis system developed by the predecessors uses a multi-step analysis strategy based on 2D image reconstruction. This strategy usually includes the following steps: vessel path detection, 2D image reconstruction, and quantification of vessel diameter based on 2D images. The basic idea of ​​the traditional 2D image reconstruction strategy is to simulate the process of manually measuring the pipe dia...

AI Technical Summary

Problems solved by technology

[0004] The current mainstream caliber analysis system based on the 2D image reconstruction strategy mainly has the following shortcomings: 1. The automatic reconstruction of the 2D section of the blood vessel is extremely dependent on the accurate retrieval of the vessel path, and the slightest deviation of the path shape will lead to an error in the section reconstruction. Thus affecting the quantitative accuracy of pipe diameter, resulting in poor robustness

It is precisely because the reconstructed image is highly dependent on the quantitative accuracy of the vascular trajectory that previous methods often involve very complicated and cumbersome steps of vascular centerline correction

2. In the blood vessel diameter measurement step based on 2D cross-sectional images, there are usually some a priori assumptions (such as assuming that the cross-sectional shape of the blood vessel is elliptical), but these assumptions are often not true in the position of the lesion blood vessel. The experimental hypothesis will also introduce measurement error

4. Many previous studies used area, average pipe diameter length, or estimated pipe diameter values ​​based on cross-sectional area to generate pipe diameter curves. Only some works used the shortest inner diameter standard commonly used in the industry to generate pipe diameter curves.

[0006] The main limitations of the cylindrical fitting strategy are: 1. The modeling process of the geometric model is cumbersome, and the ability to adapt to different vascular structure variations is poor. The parameter fitting calculation process of the model is also more complicated

2. It is often difficult to effectively describe the shape of blood vessels with complex structures based on prior geometric assumptions (such as cylindrical assumptions), which will also affect the quantitative accuracy

It is worth pointing out here that although previous studies have used the term vascular centerline to describe the vascular path, none of these methods have strictly demonstrated from the perspective of modeling that the path sought by the algorithm is really located in the center of the local blood vessel.

Images