Spatial Normalisation of 7T Data for PD calibration

hMRI-group / hMRI-toolbox

A toolbox for quantitative MRI and in vivo histology using MRI (hMRI)

GNU General Public License v2.0

61 stars 44 forks source link

Spatial Normalisation of 7T Data for PD calibration #51

Open mfcallaghan opened 2 years ago

mfcallaghan commented 2 years ago

There is no validated pipeline for spatial alignment of data in voxel space.

[ ] What should be the segmentation input?
[ ] What algorithm would have sufficient precision?
[ ] How should voxels outside the brain, where the signal models don't apply, be handled?

SiyaMRPhy commented 1 year ago

Segmentation of Outer-supressed MT was not giving good results (Dataset used - 7T 0.6 mm iso travel head data) https://github.com/SherS2/hMRI-toolbox/blob/e0cf8a68b36cb2d667707179be628badbac5068b/hmri_create_MTProt.m#L835-L859

Solution : Used multimodal segmentation using the MTsat and T1 OLS/WLS. https://github.com/SiyaMRPhy/hMRI-toolbox/blob/4ca24849652be03e86acb67bcd4b6501455a6511/hmri_create_MTProt.m#L962-L1057

SiyaMRPhy commented 1 year ago

Bias-field correction of masked A map. This step was also giving bad segmentation results.

https://github.com/SiyaMRPhy/hMRI-toolbox/blob/e0cf8a68b36cb2d667707179be628badbac5068b/hmri_create_MTProt.m#L1086

This was coming wrong TPM priors. Masked brain does not have eye (Eye in TMP is c3). Defined a TPM for the masked Brain (4 tissues classes - c1, c2, c3 without Eyes, c4 = 1-(c1+c2+c3) ) and used that for the segmentation.

https://github.com/SiyaMRPhy/hMRI-toolbox/blob/4ca24849652be03e86acb67bcd4b6501455a6511/hmri_create_MTProt.m#L1344-L1401