Hi Moran,
I am struggling exactly with the same questions, how can the connectivity matrix representing the number of reconstructed streamlines be corrected for the different seed sizes.
As far as I have read, there is absolute no consesus how to normalize the matrix. I was recommended to do it by dividing the number of streamlines connecting two regions by the values in the waytotal file or by the sum of each row. However, other papers have done this by taking the mean volume of the two regions (Collin et al 2013 Sz Bull) or by the sum of the two volumes (van den Heuvel et al 2011 JNS).
I am therefore also interested what fsl experts would recommend, which is the most appropriate approach? I guess normalizing with the waytotal also accounts for different seed volumes in an indirect way, but normalizing with the mean volume of the two regions probably addresses volume differences more directly. If this is true, can anybody help me how I can normalize the fdt_network_matrix by the mean volume easily in matlab?
Thank you very much
André
________________________________________
Von: FSL - FMRIB's Software Library [[log in to unmask]]" im Auftrag von "Moran Artzi [[log in to unmask]]
Gesendet: Samstag, 29. August 2015 15:43
An: [log in to unmask]
Betreff: [FSL] Normalization of connectivity matrix
I'm running structural connectivity matrix for a group of subject using FSL Fdt tool.
For the seed mask I used the subject's Freeserfer anatomical masks, realigned to the subjects DTI space, resulted with a bit different # of voxels in each individual seed.
Therefore in order to compare between the subjects connectivity matrix's, I first need to normalize them
What is the right way to do it?
Meany thanks
Moran
|