Hi Medani,

Maybe this discussion can be helpful for you: https://www.mathworks.com/matlabcentral/answers/380630-how-can-i-segment-mri-to-generate-a-3d-model

Hi Medani Takfarinas not sure if already exists, but you can create a pipeline using FreeSurfer and FSL libraries from Matlab. If you don't need to use MatLab, you can work directly with FreeSurfer and FSL, they are useful to generate WM, GM, and CSF volumes, not sure about Bone Segmentation using MRI images. All these processes are automatic and some papers have described that there are no statistically significant differences between fully automatic process and semi-automatic process with quality control check.

Thank you Cosimo Ligorio & Juan Felipe Orejuela Zapata for your answers.

Both of SPM and Freesurfer are good, however in order to have clean models, manual checking is needed.

As far as I know, FreeSurfer is working only on linux OS.

DUring my research I found this :

https://www.parralab.org/roast/

It's a matlab toolbox and for now it's the only pipeline that I found.

They use SPM for segmentation and iso2mesh for surface/volume mesh generation. However, I tested some scripts, it seem to be working with highly clean T1 MRI (mainly the included file the MNI152_T1) but I think not for most of available and regular MRI data.

Lets keep going and see if someone has other way.

Hi Medani,

I would suggest the check the combination spm12+cat that according to some recent publications, this combination can provide an accurate result compared to both FSL and Freesurfer. Both ROAST and Simnibs provide surface-based segmentations where each compartment should contain closed regions. On the one hand, this is a necessary step if you want to construct a tet-mesh otherwise the meshing tool will fail. On the other side, this way of segmentation/meshing is unrealistic due to the smoothing operator that causes overestimations of the original compartments (especially of the CSF compartment in the occipital and temporal brain regions).

Best,

Marios