Whole brain automate segmentation

Hello . i am using two mri PD images of one patient in 2 consecutive year of his disease.
i done below steps :

  1. i registered two images by General Registration Module ( fixed image was the image of patient in 2012 and moving image was image of patient in 2014)
  2. i applied Median Filter and i used from this Module for smoothing my images.
  3. i removed skull of brain by Swiss Skull Remove Module.
    and now i want to segment whole brain automaticlly specially GM and WM and Basal Ganglia. how should i do this?
    more explain: i used from segment editor module for segmentation but it’s results weren’t satisfactory.

I suggest SynthSeg. I’ve had good luck with it on a wide variety of scans. There’s no Slicer extension for it (yet?) but you can run it from the command line and use the results in Slicer.

Thanks for your guidness. And how about FSL software? not freesurfer

FSL has good tools too, but I haven’t used them myself.

1 Like

And i think i can import my data from FreeSurfer by FreeSurfer Importer module.right?

Yes, the SlicerFreeSurfer extension should work well. But traditional FreeSurfer requires high resolution T1 data, where SynthSeg (which comes from many of the same developers) is robust for different resolutions and contrasts.


I’ve had some success with Atropos from ANTS (An open source multivariate framework for n-tissue segmentation with evaluation on public data - PubMed).

The segmentation tool from FSL is called FAST (FAST - FslWiki).

If you choose FreeSurfer I suggest looking into FastSurfer (GitHub - Deep-MI/FastSurfer: PyTorch implementation of FastSurferCNN). segmentation finishes in half the time.


thanks. i segmented my images in 6 chapter by fsl. now i want to crop a region very accurately with high accuracy. with which tool in 3dslicer can i do this?
should i do this automaticlly?

You can use “Crop volume” module crop with an axis-aligned rectangular prism shaped region. You can clip the image using freehand shapes using Segment Editor module. Both performs pixel-perfect cropping: nothing is changed in the image except the cropped regions are removed or blanked out.