Magnetic resonance delayed enhancing sequence myocardium segmentation method

Abstract

The invention relates to a magnetic resonance delayed enhancing sequence myocardium segmentation method which comprises a first step of building a group of heart maps in advance, a second step of registering the maps and the to-be-segmented delayed enhancing sequence images, a third step of initializing all parameters in a multi-component Gaussian mixture model according to the registered map information, a fourth step of calculating a posterior probability of each pixel point in each component, a fifth step of updating an estimating value of the parameters of all the component in the Gaussian mixture model, a sixth step of evolving two outlines in a coupling level set model, a seventh step of re-estimating a prior probability of each category, and an eighth step of repeating the fourth step to the seventh step until the algorithm converges, and obtaining the final segmentation result by solving the convex hull and conducting interpolation on the inner outline in the coupling level set model. Compared with the prior art, the magnetic resonance delayed enhancing sequence myocardium segmentation method has the advantages of being convenient to realize and flexible in application without depending on other sequence information.

Description

technical field [0001] The invention relates to the technical field of image processing, in particular to a method for myocardial segmentation of magnetic resonance delayed enhancement sequence based on multi-component Gaussian mixture model and coupling level set. Background technique [0002] With the development of nuclear magnetic resonance technology, its imaging quality and imaging speed of the heart have been greatly improved, and it has been more and more widely used in relevant medical clinical practice and research. Among them, the delay enhancement sequence can enhance the signal of the myocardial infarction area by injecting contrast agent into the myocardial infarction patient, and the location and size of the infarction area can be seen intuitively, and its reliability has been verified. In order to quantitatively analyze the myocardial activity of a patient by using delayed enhancement sequences and formulate a corresponding treatment plan, it is usually first...

AI Technical Summary

Problems solved by technology

[0004] 1) Due to the existence of the infarct area, the gray distribution of the myocardium is uneven, and many segmentation algorithms are constructed based on the assumption that the target area has a uniform gray scale;

[0005] 2) There are great uncertainties in the grayscale, size, and distribution of the myocardial infarction area, and any prior assumptions may make the algorithm lose generality;

Most of these algorithms have two types of problems: First, although a lot of research has been done on myocardial segmentation of movie sequences, as prior information, any segmentation errors will be introduced into the subsequent segmentation process, so the segmentation process is cumbersome, usually human supervision required

The second is that the optimization model is relatively rough or has certain prior assumptions added, which is difficult to adapt to various complex situations of myocardial infarction patients

Images