Hello,
I am using the Nilearn function nilearn.masking.intersect_masks to create the union or intersection of two binarized masks.
According to the Nilearn documentation, the threshold argument can be set to go from intersection of the masks (threshold = 1) to union of the masks (threshold = 0).
threshold: float, optional
Gives the level of the intersection, must be within [0, 1]. threshold=1 corresponds to keeping the intersection of all masks, whereas threshold=0 is the union of all masks.
In my understanding union means an additive combination of the masks, i.e., the union contains all voxels in mask A and mask B combined. In contrast, I understand intersection as selecting only those voxels that are in both mask A and mask B but not the voxels that are only in mask A or mask B.
Alright, I did combine the masks using both thresholds and the results are a bit puzzling for me:
Okay, so from the first two images you can see that the two masks are largely overlapping (the first mask is colored in yellow and the second mask is colored in blue):
Now, if use masking.intersect_maks with threshold = 0 to (supposedly) create the union of the masks.
Well, for me this does not seem like the union of the masks (intersected mask shown in white):
The intersection (threshold = 1) does not really look like the intersection either:
So, can someone tell me what’s actually happening or could be the issue?
System specification:
- macOS Sierra version 10.12.6
nilearn==0.6.0- FSLeyes version 0.26.1+build0
Thanks in advance!