在文章中看到以下流程，如果使用freesurfer分割得到的 aparc+aseg.mgz，怎么使用这个分区去得到每个人个体空间的时间序列呢？因为bold是一个4D文件，是不是需要一个平均的EPI去进行配准。还有一个疑问是，如果用T1配准到EPI空间，那是用freesurfer得到的 brain.mgz去配准吗？
To extract regional fMRI time series from subject-specific anatomical
parcellations, each subject’s anatomical and functional data were
aligned and resampled to a common space. First, a
subject-specific brain mask, derived from the anatomical T1-weighted
image was co-registered to the EPI space of the resting-state fMRI. The
resulting transformation parameters were saved for subsequent use.
Then, the Lausanne 2018 parcellation was aligned to the EPI space
using the transformation parameters. The output parcellation was
then resampled to match the resolution and grid of the EPI data.
Finally, the transformed and resampled atlas was used as an ROI mask
to extract mean time series from each region.
是的，你的理解和我是一致的。先将BOLD的平均图像（或者第一个图像也行）配准到T1图像（如果是FreeSurfer处理过的，就对齐到brain.mgz或者nu.mgz），然后将T1空间的分区转换到BOLD空间提取时间序列。你可以使用FreeSurfer的bbregister来进行T1-BOLD的配准，是目前公认最准确的做法。我以前写过一个博客来介绍boundary-based registration，但是用的是FSL，效果肯定是不如直接用bbregister的。