Dear MRtrix experts,
When using “5ttgen hsvs”, I encountered unexpected results in some of the subjects. In these subjects, the 5ttgen result are something like this:
in which most of the voxels are 0.
A closer look into the intermediate files of 5ttgen showed that the problem seemed to be related to the process of converting smoothed freesurfer’s vessel segmentation from a mesh surface to a volume image. For example:
Before smoothing:
Left-vessel_init.vtk
# vtk DataFile Version 1.0
ASCII
DATASET POLYDATA
POINTS 24 float
-27.069499350611892 6.9711386462253842 -23.84210967936076
-27.569499406997494 7.4711387100615951 -23.842109679244345
-27.069499347378297 7.4711387058300289 -24.342109679244345
-26.569499287788204 7.4711387015984627 -23.842109679244345
-27.069499344173806 7.9711387654346737 -23.84210967912793
-26.069499224964517 7.9711387569715413 -23.84210967912793
-26.569499281350119 8.4711388208077523 -23.842109679011514
-26.069499221730922 8.4711388165761861 -24.342109679011514
-25.569499162140829 8.4711388123446199 -23.842109679011514
-26.069499218526431 8.9711388761808308 -23.842109678895099
-25.069499099317142 8.9711388677176984 -23.842109678895099
-25.569499155702744 9.4711389315539094 -23.842109678778684
-25.069499096083547 9.4711389273223432 -24.342109678778684
-24.569499036493454 9.471138923090777 -23.842109678778684
-25.069499092879056 9.971138986926988 -23.842109678662268
-22.069498728813102 10.97113908074688 -23.842109678429438
-22.569498785198704 11.471139144583091 -23.842109678313022
-22.069498725579507 11.471139140351525 -24.342109678313022
-21.569498665989414 11.471139136119959 -23.842109678313022
-22.069498722375016 11.97113919995617 -23.842109678196607
-27.069499347407401 7.4711387058300289 -23.342109679244345
-26.069499221760026 8.4711388165761861 -23.342109679011514
-25.069499096112651 9.4711389273223432 -23.342109678778684
-22.069498725608611 11.471139140351525 -23.342109678313022
POLYGONS 32 128
3 2 0 1
3 0 2 3
3 2 1 4
3 5 6 7
3 3 2 4
3 5 7 8
3 7 6 9
3 10 11 12
3 8 7 9
3 10 12 13
3 12 11 14
3 13 12 14
3 17 15 16
3 15 17 18
3 17 16 19
3 18 17 19
3 0 20 1
3 0 3 20
3 1 20 4
3 3 4 20
3 5 21 6
3 5 8 21
3 6 21 9
3 8 9 21
3 10 22 11
3 10 13 22
3 11 22 14
3 13 14 22
3 15 23 16
3 15 18 23
3 16 23 19
3 18 19 23
When using mesh2voxel to convert “Left-vessel_init.vtk” to volume
mesh2voxel Left-vessel_init.vtk aparc.mif Left-vessel_init.mif
, the resulting “Left-vessel_init.mif” seemed to be OK.
After smoothing:
Left-vessel.vtk
# vtk DataFile Version 1.0
ASCII
DATASET POLYDATA
POINTS 24 float
-27.069499349165412 7.1375275154699915 -23.842109679365393
-27.403110537753289 7.4711387090282875 -23.842109679238824
-27.069499347377434 7.4711387057864371 -24.17572071075719
-26.735888157032054 7.4711387026313272 -23.842109679249866
-27.069499345619938 7.8047498961896258 -23.842109679123304
-26.069499223519013 8.1375276262147729 -23.842109679132516
-26.403110412105466 8.4711388197745769 -23.842109679005812
-26.069499221730769 8.4711388165328625 -24.175720710524097
-25.735888031386896 8.4711388133777952 -23.842109679017216
-26.069499219973366 8.8047500069375992 -23.842109678890512
-25.069499097871631 9.1375277369612835 -23.842109678899508
-25.403110286458624 9.4711389305205564 -23.842109678772449
-25.069499096082858 9.4711389272790178 -24.175720710290733
-24.735887905738636 9.4711389241241317 -23.842109678784563
-25.069499094325632 9.8047501176834029 -23.842109678657511
-22.069498727366614 11.137527949991352 -23.842109678433758
-22.40310991595388 11.471139143549829 -23.842109678307501
-22.069498725578818 11.471139140308111 -24.175720709826141
-21.735887535233886 11.471139137152869 -23.842109678318188
-22.069498723821148 11.804750330711343 -23.842109678191932
-27.069499347407902 7.4711387058731757 -23.508498647731493
-26.0694992217616 8.4711388166195114 -23.508498647498936
-25.069499096114395 9.4711389273656685 -23.508498647266279
-22.069498725608941 11.471139140394586 -23.508498646799552
POLYGONS 32 128
3 2 0 1
3 0 2 3
3 2 1 4
3 5 6 7
3 3 2 4
3 5 7 8
3 7 6 9
3 10 11 12
3 8 7 9
3 10 12 13
3 12 11 14
3 13 12 14
3 17 15 16
3 15 17 18
3 17 16 19
3 18 17 19
3 0 20 1
3 0 3 20
3 1 20 4
3 3 4 20
3 5 21 6
3 5 8 21
3 6 21 9
3 8 9 21
3 10 22 11
3 10 13 22
3 11 22 14
3 13 14 22
3 15 23 16
3 15 18 23
3 16 23 19
3 18 19 23
When using mesh2voxel to convert “Left-vessel.vtk” to volume
mesh2voxel Left-vessel.vtk aparc.mif Left-vessel.mif
, the resulting “Left-vessel.mif” seemed to be weird:
In “Left-vessel.mif”, most of the voxels were 1, while voxels corresponding to vessels were of lower value (around 0.5).
Any hints to why this happened ?
Many thanks,
Scimon