Skip to content. | Skip to navigation

Sections
Personal tools
You are here: Home → Forum → Using data sets → vim-1: indexing method and category information?
 

vim-1: indexing method and category information?

Up to Using data sets

vim-1: indexing method and category information?

Posted by Xixi Wang at August 18. 2015

Hi all, I'm trying to analyze the vim-1 data and then perform category based similarity analysis, here are two questions I got:

1) converting the voxIdx<S1/S2> to image coordinates(i, j, k):

I tried the ind2sub function in MATLAB - to extract image coordinates for V1 region, I used: [i, j, k] = ind2sub([64, 64, 18], voxIdx(roiS1 == 1)) and then I generated the corresponding mask for V1. But when I overlayed the mask on the bold data, the V1 region is wrong. My guess is that for MATLAB, the indexing starts from the top slice, but for the BOLD data, since it's in LPI system, the indexing starts from the bottom slice. So I wonder if it's the case and what's the correct way to convert voxIdx to ijk coordinates.

2) I'm trying to calculate the similarity between brain patterns of different categories, so I wonder if there is any category information of the image stimuli for each run. 

Best,

Xixi

Re: vim-1: indexing method and category information?

Posted by admin at September 04. 2015

Response below by Kendrick Kay:

Hi Xixi

1) I wanted to generate different ROI masks in MATLAB so I tried converting the voxIdx<S1/S2> to image coordinates(i, j, k):
i.e. for V1: [i, j, k] = ind2sub([64, 64, 18], voxIdx(roiS1 == 1))
I then wrote the extracted matrices to NIFTI files. But when I overlayed the mask on the bold data, the V1 region was wrong. My guess is that for MATLAB, the indexing starts from the top slice, but for the BOLD data, since it's in LPI system, the indexing starts from the bottom slice. So I wonder if that's the case and what's the correct way to convert voxIdx to ijk coordinates.
 
It might be problem in writing the NIFTI files.  The indexing is just standard MATLAB indexing, so i think what you wrote is correct.  i should be the index in the first dimension (rows), j should be the second dimension (cols), k should be the slice dimension (3rd dimension).
 
2) I'm trying to calculate the similarity between brain patterns of different categories. To avoid manually classifying those image stimuli, I wonder if there is any category information of the stimuli for each run.
No, not that i'm aware of.  You can try contacting Thomas Naselaris <[email protected]> who might have some information.
 
 
Kendrick

 

 

 

Powered by Ploneboard
Document Actions