Nilearn result region output and display question

sarah20 · February 15, 2022, 8:16pm

Hello,

Recently I had asked the Nilearn team for help putting in coordinates and getting the region from the DiFuMo atlas from those coordinates and they were kind enough to write that language, so here is a link to that:

gist.github.com

https://gist.github.com/bthirion/61630ace721c639e16f459ea212b3e94

position_label.py

from nilearn.datasets import fetch_atlas_difumo
from nilearn.input_data import NiftiSpheresMasker
import numpy as np

d256 = fetch_atlas_difumo(dimension=256)
seeds = [[-52, 14, 0]] # this is your position, you may pick any you wish
masker = NiftiSpheresMasker(seeds=seeds, radius=2).fit()
weights = masker.transform(d256.maps)
label = d256.labels[np.argmax(weights)]
print('label:', label)

Relatedly, when depicting a statistics image from results, does Nilearn have the ability to depict both an ICA type image with something like a t-map overlayed within the same command? So, in effect including two results within one command. FSLeyes allows this in their viewer, but I would like to use Nilearn if possible. Thank you!

bthirion · February 15, 2022, 9:30pm

Do you mean that you want to overlay a statistical image on top of an ROI image ? Or that you want to display simultaneously two real-valued images ?
Best,
Bertrand

(Note that the above script is super slow… depending on your need, it can probably be made way more efficient.)

sarah20 · February 15, 2022, 9:53pm

For the ICA image, I have the option to show either what appears to be just a 0/1 mask of the component, or I think a z-stat image. I want to overlay a t-stat image on that to show a significant cluster that covaries with the ICA network. Is that possible with Nilearn? Thank you!