基于图像非线性配准的MRI几何失真校正

采用图像非线性配准的方法,对MRI存在的几何失真进行校正,先选取与MRI具有共同解剖结构且基本无几何失真的CT图像,作为配准校正的目标基准图像,将MRI与CT图像进行刚体配准,然后采用人工选点的方法,寻找CT图像和MRI中的对应标记点集,用改进的弹体样条函数进行全局非线性配准.结果表明:弹体样条变换具有较强的非线性校正能力,较好地实现了二维MRI的几何失真校正.我们采用的变换模型有效可行,并可用于三维失真图像的校正.

Correction of Geometric Distortion in MRI Based on Non-linear Image Registration

To correct the geometric distortion by means of non-linear image registration.CT image was selected as reference image,which has the same anatomical structure as MRI but almost has no geometric distortion.MRI was matched with CT image by rigid body registration and a manual approach was used to seek corresponding points between CT image and MRI.Then,the improved elastic body spline was adopted to make global non-linear registration with these corresponding points extracted.Results showed that the elastic body spline transformation possessed strong ability to correct non-linear distortion and a desirable 2-D corrected MRI was got.Experiments show that it is an effective correction approach and can be conveniently used to the correction of 3-D distorted MRI.