Importing scalar image and its binary labelmap in NIFTI format

[AbdulrahmanMohammed2021]

Hi Fernando,
I've imported a MRI scan of head and its binary labelmap as a volume and as segmentation, respectively. However, the output volume didn't align with segmentation. So can you please give any explanation for that?

Note: I'm wanna to save the augmented volume and its segmentation(binary labelmap) as actual augmentation in NIFTI format. I'm using MATLAB so that the reason why I can't perform augmentation during training.

Thanks

[fepegar]

Hi, @AbdulrahmanMohammed2021. This is not related to TorchIO. You probably have different spatial metadata for both images.

[AbdulrahmanMohammed2021]

Hi,@fernando, just to let you know that I've done the segmentation on Slicer, and exported images and their segmentation into NIFTI files. Then in another Slicer scene, I imported them ( scalar image as a volume and volume labelmap as segmentation) to perform augmentation using TorchIO. When I applied the augmentation, it is only applied to the scalar image and It doesn't align with the labelmap as shown in the previous post. Can you elaborate on that?

I hope it is clear

thanks

[fepegar]

Thanks for clarifying. Although TorchIO performs the same operation to all images in the Subject, the Slicer module doesn't have support for that. That is because it is conceived to play with the transforms parameters, not to perform the actual augmentation for training. If you use the snippet I posted on the Slicer forum, your transformed images will be correctly aligned.

[AbdulrahmanMohammed2021]

OK, that's make sense I've tried the snippet and as you said, they were aligned correctly. last question, can I save the transformed images in NIFTI format?

thanks

[fepegar]

Sure! You can use the save() method of the images.