[BUG] RegionExtractor does not have a two-tailed behavior.

[bthirion]

Is there an existing issue for this?

• I have searched the existing issues

Operating system

• Linux
• Mac
• Windows

Operating system version

For example one of the following:

• Linux Ubuntu 22.04
• Mac OS Version 12 "monterey"
• Windows 11

Python version

• 3.12
• 3.11
• 3.10
• 3.9
• 3.8

nilearn version

0.10.2

Expected behavior

Region Extractor should include the possibility to extract only supra-threshold regions (on-sided behavior) --- an this should be the default IMHO. Currently it has a only the two-tailed behavior, meaning that tit won't distinguish between positive and negative regions.

Current behavior & error messages

This is what I got:

I would like to have

Steps and code to reproduce bug

from nilearn.regions import RegionExtractor
from nilearn import datasets
from nilearn.image import resample_to_img
from nilearn.plotting import plot_prob_atlas, plot_stat_map, show
from nilearn.image import concat_imgs

image = datasets.load_sample_motor_activation_image()
mask_img = datasets.load_mni152_gm_mask()
mask_img_ = resample_to_img(mask_img, image)
image = concat_imgs([image])
threshold = 0.0
 
extractor = RegionExtractor(
	maps_img=image,
	mask_img=mask_img_,
	min_region_size=1000,
	threshold=threshold,
	thresholding_strategy='img_value',
	extractor='connected_components', )

# Just call fit() to execute region extration procedure
coords = (15, -27, 46)
extractor.fit()
regions_img = extractor.regions_img_
plot_prob_atlas(regions_img, cut_coords=coords)
plot_stat_map(image, threshold=threshold, cut_coords=coords)

show()