weights.csv, meanlength.csv and connectome.csv

[zhangxiaobai12345]

Hi! I'm sorry to bothor you! I've run participant level and the get these three .csv .
I wonder the meaning of them, does weights.csv contains a fiber-weight matrix, which shape is n*n, n is the number of regions? However, it seems about 300M, which is too large for data, as the meanlength.csv is 7kb and connectome.csv is 522kb.
Actually, I wonder the difference between meanlength.csv and connectome.csv.

My instruction is:
python mrtrix3_connectome.py BIDS/sub-01/output/MRtrix3_connectome-preproc/ BIDS/sub-01/output participant -parcellation brainnetome246mni -atlas_path data/tvb_optimize_sc/ -output_verbosity 4 -force

[Lestropie]

weights.csv should contain one floating-point value per streamline. While it might look like 2D matrix data in a text editor if line-wrapping is enabled, it's actually just one long list. It can be quite large if the tractogram contains a large number of streamlines; that fact that it's encoded as text rather than binary data makes it larger again, but I've kept it that way for simplicity.

connectome.csv encodes within each edge the sum of streamline weights, which is a proportional estimate of the total intra-axonal cross-sectional area of the WM connections between those two endpoints. With use of the group analysis level, this becomes the Fibre Bundle Capacity (FBC) measure. See relevant description. meanlength.csv is instead simply the mean length in mm of each edge (actually a weighted mean, where the contribution of each individual streamline length to that mean is modulated by its weight as calculated by SIFT2).

[zhangxiaobai12345]

weights.csv should contain one floating-point value per streamline. While it might look like 2D matrix data in a text editor if line-wrapping is enabled, it's actually just one long list. It can be quite large if the tractogram contains a large number of streamlines; that fact that it's encoded as text rather than binary data makes it larger again, but I've kept it that way for simplicity.

connectome.csv encodes within each edge the sum of streamline weights, which is a proportional estimate of the total intra-axonal cross-sectional area of the WM connections between those two endpoints. With use of the group analysis level, this becomes the Fibre Bundle Capacity (FBC) measure. See relevant description. meanlength.csv is instead simply the mean length in mm of each edge (actually a weighted mean, where the contribution of each individual streamline length to that mean is modulated by its weight as calculated by SIFT2).

Thanks for your reply!!!

[zhangxiaobai12345]

weights.csv should contain one floating-point value per streamline. While it might look like 2D matrix data in a text editor if line-wrapping is enabled, it's actually just one long list. It can be quite large if the tractogram contains a large number of streamlines; that fact that it's encoded as text rather than binary data makes it larger again, but I've kept it that way for simplicity.

connectome.csv encodes within each edge the sum of streamline weights, which is a proportional estimate of the total intra-axonal cross-sectional area of the WM connections between those two endpoints. With use of the group analysis level, this becomes the Fibre Bundle Capacity (FBC) measure. See relevant description. meanlength.csv is instead simply the mean length in mm of each edge (actually a weighted mean, where the contribution of each individual streamline length to that mean is modulated by its weight as calculated by SIFT2).

Hi! Sorry to bothor you! I've use the connectome.csv and meanlength.csv to run FAST-TVB, however get a correlation of 0.02 with the empiral FC, it seems that the strength is too weak, about 1e-7+1 with G = 0.3, how can I improve it?