Dana and David Dornsife Cognitive Neuroscience Imaging Center, University of Southern California, Los Angeles, California, USA.
J Magn Reson Imaging. 2013 Jun;37(6):1460-7. doi: 10.1002/jmri.23929. Epub 2012 Nov 21.
To correct distortions caused by eddy currents induced by large diffusion gradients during high angular resolution diffusion imaging without any auxiliary reference scans.
Image distortion parameters were obtained by image coregistration, performed only between diffusion-weighted images with close diffusion gradient orientations. A linear model that describes distortion parameters (translation, scale, and shear) as a function of diffusion gradient directions was numerically computed to allow individualized distortion correction for every diffusion-weighted image.
The assumptions of the algorithm were successfully verified in a series of experiments on phantom and human scans. Application of the proposed algorithm in high angular resolution diffusion images markedly reduced eddy current distortions when compared to results obtained with previously published methods.
The method can correct eddy current artifacts in the high angular resolution diffusion images, and it avoids the problematic procedure of cross-correlating images with significantly different contrasts resulting from very different gradient orientations or strengths.
在高角分辨率扩散成像中，纠正由于大扩散梯度引起的涡流导致的失真，而无需任何辅助参考扫描。
通过仅在扩散梯度方向接近的扩散加权图像之间进行图像配准，获得图像失真参数。数值计算了一个线性模型，该模型将失真参数（平移、比例和剪切）描述为扩散梯度方向的函数，以允许对每个扩散加权图像进行个性化失真校正。
该算法的假设在一系列对体模和人体扫描的实验中得到了成功验证。与以前发表的方法相比，在所提出的算法应用于高角分辨率扩散图像时，明显减少了涡流失真。
该方法可以校正高角分辨率扩散图像中的涡流伪影，并且避免了由于梯度方向或强度非常不同而导致对比度明显不同的图像进行交叉相关的有问题的过程。
