You might want to have a look here, too:

http://humanconnectome.org/about/pressroom/project-news/announcing-release-of-s900-ptn-and-other-group-average-data/

Specifically the HCP_S900 group average functional connectivity.

In addition to the above suggestions, you could consider the template proposed in Jonathan Power's 2011 paper (http://www.ncbi.nlm.nih.gov/pubmed/22099467). See attached template details and NifTi image. Before usage, you should coregister the template to your data using nearest neighbor interpolation.

Would anyone happen to have the templates for salience and executive networks based on Power's 2011 paper?

Many thanks in advance!

Hi Lucia. Find attached template for salience network, with ROI spheres of radius 5mm around each ROI. Before usage, you should coregister the template to your data using nearest neighbor interpolation.

Regarding executive network, which exactly are you referring to? The Cingulo-opercular Task Control network or the Fronto-parietal Task Control network?

Hi Ruedeerat. Find attached Power et al. template for all ROIs from the entire brain.

Hi Cory, all the ROIs are spheres of radius 10mm.

To create these ROIs, I obtained the Power et al. 2011, Neuron, ROI centroids from the author's website: http://www.nil.wustl.edu/labs/petersen/Resources.html

The ROI spheres were created using Marsbar.

Yes, overlapped ROIs are certainly a possibility because the ROIs were drawn as 10mm-radii spheres around the Power et al. atlas centroids. Wherever the ROI centroids are within 20mm apart, there would be an overlap. As I understand, all ROIs have the same size (in voxels). If a gray matter mask is enforced, the number of voxels would change, but it would encompass all gray matter within 10mm of the ROI centroid.

Hi Rangaprakash, 

Thanks for sharing these ROIs. Could you also share the script you've used to generate them? Thanks!

Sean

Hi Sean,

Sure, see attached. The readme file has the necessary info to create the ROIs. All the best!

-Ranga

Hi Rangaprakash, 

I have a follow up question on how you classified the ROI nodes into anatomical labels. Can you share your procedures? Thank you again!

Sean

Hi Sean,

Actually I did not derive or make the Power264 template, hence the anatomical labels are defined by the creators of this template, i.e. Power et. al. 2011, Neuron.

The ROI centroids were taken from the author's website: http://www.nil.wustl.edu/labs/petersen/Resources.html

-Ranga

Hi Rangaprakash D. ,

I am not able to use the get_marsbar_rois scripts since there are some Windows specific functions used and I am using Linux. Can you please help me with the .nii files for PP264_all_ROIs_combined of radius 4mm and 2x2x2 voxels?

Thanks!

Aagam

Hi Aagam, the scripts are in the Matlab platform and it should work irrespective of the OS. My friends have been able to run it on Mac. Nevertheless, you could take the existing " PP264_all_ROIs_combined.nii" image and co-register it with your 4mm image using nearest neighbor interpolation to obtain a mask for your resolution and reference space.

Dear Rangaprakash,

I have few doubts related to region based analysis and small volume correction (SVC). Since it is related to creating mask I am posting this question here.

As far I know SVC performs voxel-wise testing. If I make sphere of 12 mm size and use it to get final output result, will it be the region based (considering entire region as one voxel / mean) analysis or voxel-wise analysis. Suppose if I create a mask of Hippocampus apply the mask to get the result, will it be region based or voxel based analysis?

Please clarify.

Thanks in advance

RB

Hi,

I have noticed that in the original paper they talk about spheres of 10mm, you derived spheres with 5mm diameter which solves the overlapping issue. However, the ROIs where generated from functional patterns what does entitle changing the size of the ROIs? I mean is it a sensitive thing? Or we are just solving the overlap issue and introduce imprecision?

Hi Alessandro,

The choice of radius is not clear in any literature. The choice of 10mm in one of the original papers seems arbitrary. The value '10' is a neat rounded figure; apart from that I see no rationale for choosing 10mm over 8mm or 11mm. I chose 5mm to reduce the overlapping issue - at least choosing 5mm solves a problem. When all choices are equally arbitrary, why not choose the one that's better. That's my argument.