Brain map and brain image registration method based on high-order statistic deformable model

Abstract

The invention provides a brain map and brain image registration method based on a high-order statistic deformable model. The brain map and brain image registration method based on the high-order statistic deformable model comprises a first step of selecting a three-dimensional brain map I, selecting N three-dimensional images Mi as training samples, using an affine registration method for registering Mi to the three-dimensional brain map I to obtain MA i , and then using a non-rigid body registration method for registering the images MA i to the three-dimensional brain map I to obtain a series of deformation field vectors fi, a second step of forming four-order tensors Ai by the deformation field vectors fi, solving the mean value A(_) of the four-order tensors Ai, setting Alpha 1= Ai-A(_), and using a lower dimension four-order kernel tensor and four basis matrixes for expressing an estimated value Alpha(^)1 of the Alpha 1, conducting minimization to obtain the optimal solution of the basis matrixes, and expressing the tensors Ai formed by the deformation field vectors through a low dimension four-order kernel tensor and the optimal solution, a third step of conducting deformation on the image needed to be registered based on the obtained Ai and obtaining SD and a fourth step of using the non-rigid registration method for registering the image SD to the brain map I.

Description

technical field [0001] The invention relates to the technical field of medical image processing, in particular to a brain atlas and brain image registration method based on a high-order statistical deformation model. Background technique [0002] People have been able to obtain high-dimensional images of the anatomy and function of the brain, which have brought revolutionary changes to clinical diagnosis, surgical planning and guidance, and disease treatment. In these applications, it is very important to determine the important region, that is, where the region of interest is located in the image. In the past, doctors usually judged the region of interest from anatomy books, atlases and their own experience. Even experienced doctors had difficulty connecting with actual images of patients, let alone doctors who lacked clinical experience. The digital brain atlas can solve this problem very well. [0003] The digital brain atlas is the result of 3D segmentation processing ...

AI Technical Summary

Problems solved by technology

The current registration algorithms contain a lot of parameters. In some cases, a mismatch will occur when using the default parameters. At this time, the parameters need to be adjusted. The general registration algorithm takes a long time. When this happens, Relevant parameters need to be adjusted many times, which will bring a lot of inconvenience to use, especially not suitable for clinical application

Images