How to use the publicly shared fMRI data processed by C-PAC pipeline?

[Erick Zhou]

Dear All

 

I hope this email finds you well. I have a question about how to use the publicly shared fMRI data by C-PAC pipeline.

From the ABIDE website, I downloaded the ABIDE I Preprocessed data by C-PAC pipeline that has the dimension of <61×73×61×T> (T is the number of time points). I also downloaded the Harvard-Oxford (HO) atlas, which has a dimension of <65×77×63>. The fMRI data and the HO atlas have the same voxel size of 3mm×3mm×3mm. Does the C-PAC pipeline will lead to different dimensions of atlas and the fMRI data? How can I get voxel time series from the preprocessed fMRI data of ABIDE I? Thank you.

 

Best Regards

Zhongxing Zhou

[sgia...@gmail.com]

Hello,

So you already have fMRI data that has been preprocessed by C-PAC, and you want to extract the voxel time series using the atlas that has different dimensions, correct?

Usually C-PAC does this automatically during a run if you provide the filepath to the atlas in the time series extraction section in the pipeline configuration. It will either realign the atlas to the fMRI data, or realign the fMRI data to the atlas (based on the user's decision), and then it runs the extraction.

In your case, since the data has already been pre-processed, this can be done using some tools:

• FSL flirt (https://fsl.fmrib.ox.ac.uk/fsl/fslwiki/FLIRT/UserGuide#flirt) can be used for the simple realignment/resampling.
  • Note this is not a perfect solution: realigning the fMRI data to the atlas will interpolate the fMRI data, blurring it slightly (and will also take awhile if it has many TRs) - and realigning the atlas to the fMRI data might distort the boundaries of the atlas, especially if certain ROI regions are very small.
  • It's important to use nearest neighbor interpolation if you are realigning the atlas to the fMRI data, as using another interpolation will erroneously introduce non-binary values to the atlas.
• AFNI 3dROIstats (https://afni.nimh.nih.gov/pub/dist/doc/program_help/3dROIstats.html) can be used to extract the mean voxel intensity value for all voxels within each ROI in the atlas.

I hope this helps, let me know if you have any questions.

Best,

Steven Giavasis

Lead Research Engineer

Computational Neuroimaging Lab

Child Mind Institute

childmind.org

[Erick Zhou]

Dear Mr. Steven Giavasis,  

 

I am very grateful for your answer I have waited for a month to get answers.  

 

The tools you provided may be not suitable for my situation. In my case, the HO atlas and processed fMRI data have the same voxel size. The difference between them is the number of planes of zeros. The HO atlas  has more planes of zeros. I try to match the origin of the HO atlas and processed fMRI data, and remove some planes of zeros from the  

HO atlas. Then I got the voxel time series. To make sure if the obtained voxel time series is correct. I calculated the ROI time series by using the HO atlas and the voxel time series. Then I compared the calculated ROI time series with the ROI time series directly from the ABIDE I dataset. But the results are not matched. The details can be found in the attachments of this email. Would you please read the attachments at your convenient time? Thank you. 

 

Best Regards 

Zhongxing Zhou 

 Test_Data.rar

sgia...@gmail.com <sgia...@gmail.com> 于2021年1月8日周五 上午7:16写道：

--
You received this message because you are subscribed to the Google Groups "cpax_forum" group.
To unsubscribe from this group and stop receiving emails from it, send an email to cpax_forum+...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/cpax_forum/7dc47984-23a4-41ca-be1d-eb46e1feff7fn%40googlegroups.com.