Method and system for matching MR image feature points before and after nonlinear deformation of biological tissue

Abstract

The present invention relates to a method and a system for matching MR image feature points before and after the nonlinear deformation of a biological tissue. According to the technical scheme of the invention, a feature point automatic detection method based on a depth-cascaded convolutional neural network is provided. According to the method, firstly, a general region of feature points is obtained through the first layer of the depth convolutional network. Secondly, the position of a target feature point is approximated step by step in the second and third layers of the cascade convolutional network, so that the detection rate of feature points is further improved. The method aims to solve the problem in the prior art that the feature point distinguishing ability is reduced due to the image nonlinear deformation of existing feature point descriptors. In this way, a Riemannian manifold is combined with the kernel method to construct a nonlinear deformation feature point descriptor for robustness. The three-dimensional feature points of a magnetic resonance image are mapped into a four-dimensional Riemannian manifold space. Meanwhile, the feature points are further mapped into a higher-dimensional Hilbert space based on the kernel method, so that a richer description of data distribution is obtained. Meanwhile, a real geometric distance between feature points is obtained, so that the feature points are matched.

Description

technical field [0001] The invention relates to the matching of feature points of heterogeneous biological tissues in three-dimensional nuclear magnetic resonance images, in particular to a method and system for matching feature points of MR images before and after nonlinear deformation of biological tissues. Background technique [0002] In the study of deformation field measurement of heterogeneous bodies, the key problem is how to extract a large number of feature points and how to achieve the correct matching between feature point pairs. The matching and calculation based on feature points has the advantages of fast speed and can handle large deformations. However, the correct matching of a large number of feature points has become a bottleneck restricting the high-precision measurement of deformation fields, and it is urgent to find corresponding solutions. The following describes the research status of feature point detection and description, as well as non-rigid point...

AI Technical Summary

Problems solved by technology

[0015] (1) Although feature point descriptors such as SIFT and SURF have good performance of invariance such as scale, illumination, affine transformation, perspective transformation, etc., their inherent description methods for local neighborhoods cannot solve the problem of large image problems. Characterization Problems During Deformation

However, feature point descriptors such as GIH and LGS still have problems such as too ideal description, algorithm failure in some cases, and high complexity, and in fact fail to describe the general deformation exactly.

[0016] (2) There are two main problems with the non-rigid point set matching algorithm. 1. The point set matching method mainly uses the relationship between points inside the point set. It is aimed at the feature analysis of graphics and fails to use gray images degree, texture and other information

2. Using the label relaxation method to iterate the global objective function is often easy to fall into the local extremum

In fact, in the case of a large number of feature points, there may be a small number of feature points that show anisotropy inconsistent with the overall objective function. Using a global objective for optimization will lead to wrong matching

3. In the case of the existence of mismatch probability, there is a lack of corresponding mismatch elimination strategies. In the study of volume deformation field measurement, mismatch point pairs will lead to inability to obtain accurate physical data such as elastic modulus and Poisson coefficient locally. parameter

It can be seen from the above analysis of the current research status at home and abroad that the non-linear image registration method based on grayscale and curve and surface features is essentially difficult to achieve the goal of heterogeneous tissue deformation measurement

Images