Wrote preprocessing script for reproduction of paper analysis #31
Conversation
| run_data.append(data[..., 4:]) | ||
| concatenated_run_data = np.concatenate(run_data, axis=3) | ||
| concatenated_img = nib.Nifti1Image(concatenated_run_data, | ||
| np.eye(4)) |
There was a problem hiding this comment.
maybe this is a good place to apply the gaussian filter
see "More Notes" under Lectures
from scipy.ndimage import gaussian_filter
Smooth by 2 voxel SD in all three spatial dimensions
smooth_data = gaussian_filter(data, sigma = [2, 2, 2, 0])
The sigma = [2,2,2,0] is for 2 SDs - seems standard, 95% CI
| "runs" : [ | ||
| { | ||
| "linear" : { | ||
| "path": "data/raw/sub001/task001_run001/bold_dico_dico7Tad2grpbold7Tad.nii", |
There was a problem hiding this comment.
don't we want the nonlinearly aligned version from Hanke, rather than the linearly aligned version, based on what we talked to Jarrod and Matthew about in OH? (i.e. bold_dic_dico7Tad2grpbold7Tad_nl.nii)
Is that a big deal to change? If so, it's okay b/c we should be using the rcds version anyway
There was a problem hiding this comment.
we're going to work only with rcds versions. I'll make this update in the next PR
|
Ok I think this looks fine |
Wrote preprocessing script for reproduction of paper analysis
No description provided.