Lfolds-model-fd4 and Rfolds-model-fd4

These models are for 3.1 graphs which include cortical thickness information. They may be used for morphometry, but not for sulcal recognition yet because they have not been trained on a suitable database.

These models provide descriptive figures in Talairach space when possible, which was not always what the older models (2000-2001, FD2) used to do. They are based on the "Fold Descriptors 4" generation, which complete and improve the models based on "Fold Descriptors 2/3". Descriptors which are not documented here are documented on those older models documentation.

We may provide in the future a vartiant of these models which act in native space rather than in Talairach space, which may be useful in some morphometry studies.

Sulci descriptors

index: name: description:
1. valid
2. extremitiesValid
3. extremity1x
4. extremity1y
5. extremity1z
6. extremity2x
7. extremity2y
8. extremity2z
9. gravityCenter_x
10. gravityCenter_y
11. gravityCenter_z
12. normalValid
13. normal_x
14. normal_y
15. normal_z
16. direction_x
17. direction_y
18. direction_z
19. surface Be careful that the surface now replaces the volume from older models.
20. geodesicDepthMax
21. geodesicDepthMin
22. geodesicDepthMean Mean geodesic depth of the sulcus. This feature did not exist in older models
23. connectedComponentsAllRels
24. connectedComponents
25. pureCortical
26. distanceBetweenComponentsMax
27. plisDePassage
28. hullJunctionsLength This is now the "real" geodesic length of the sulcus. It replaces the size (of voxels, in mm3) from older models.
29. GM_thickness Mean grey matter thickness in the sulcal zone (voronoi zones of nodes forming the sulcus).
30. fold_opening Fold opening index. It is calculated as the CSF volume in the fold divided by the surface of the sulcus (of the skeleton), so it indicates a mean thickness of the CSF in the fold.
31. fold_descriptor4_output Just random. Don't use it in current models !

Sulcal relations descriptors

Currently, the relations are the same as the older models.