Method and system for registering diffusion tensor images

Abstract

The invention relates to a method for registering diffusion tensor images. The method includes determining spatial neighborhoods of each pixel in each to-be-registered diffusion tensor image; extracting rotational invariance characteristics of the diffusion tensor images; measuring tensor similarities of the diffusion tensor images and reference images; registering spatial positions of the diffusion tensor images, and establishing spatial position registration objective functions; optimally solving the established spatial position registration objective functions; measuring scalar similarities of the diffusion tensor images and the reference images; registering the scalar similarities of the diffusion tensor images by the aid of a process on the basis of polynomial expansion. The invention further relates to a system for registering the diffusion tensor images. The method and the system have the advantages that tensor direction information in the diffusion tensor images is sufficiently utilized for the characteristic that the tensor direction information in the diffusion tensor images is more important that grey information, and accordingly the diffusion tensor images can be precisely registered.

Description

Technical field [0001] The invention relates to a diffusion tensor image registration method and system. Background technique [0002] Diffusion Tensor Imaging (DTI), a new method of describing brain structure, is a special form of magnetic resonance imaging (MRI). For example, if nuclear magnetic resonance imaging is to track hydrogen atoms in water molecules, then diffusion tensor imaging is based on the direction of movement of water molecules. [0003] Diffusion tensor imaging images (hereinafter referred to as "diffusion tensor images") are presented in a different way from previous images. They can reveal how brain tumors affect nerve cell connections, guide medical staff to perform brain surgery, and can also reveal the same as stroke and multiple sclerosis Minor abnormal changes related to dyslexia, schizophrenia, and dyslexia. [0004] Diffusion tensor images are different from conventional medical images. The information at each voxel is not a gray value, but a second-ord...

AI Technical Summary

Problems solved by technology

This makes the registration of diffusion tensor images very different from that of conventional scalar images: First, the direction information of the tensor in the diffusion tensor image is more sensitive than the gray level information, and the registration accuracy has far greater influence on the direction information. It is larger than the grayscale information; secondly, the similarity measure of the diffusion tensor image is difficult to define, which is different from the similarity measure of the scalar image which can be directly defined by the gray value of the image

At present, there is no more accurate method or system for diffusion tensor image registration

Images