Hi I have been trying to run fixel based analysis based on the instructions ( http://mrtrix.readthedocs.io/en/latest/workflows/fixel_based_analysis.html ) and following this article ( http://www.sciencedirect.com/science/article/pii/S1053811915004218 ) I do not understand how to proceed. I wrote foll…

Hi Mamun, The contrast_matrix.txt is an input (not a result). It’s a text file with a single row defining your contrast vector. Assuming you have patients listed first (i.e. the first 8 rows), then it should be [1 -1] to test for controls > patients. Note the input design matrix and contrast for f…

If I could take this a step further… When performing fixel-wise correlations with other metrics using fixelcfestats I understand that it behaves like FSL but need to explicitly add a column of 1s. I have two groups and an associated (de-meaned) global measure. Am I right in saying that my design m…

Hi I was trying to do something similar while including the groups as covariates so I demeaned all the columns but the first, then the design matrix would be something like (if it has only the 4 rows): 1 0.5 0 -4 1 0 0.5 1 1 0 0.5 6 1 0.5 0 -3 With the contrast as you say: 0 0 0 1 I a…

While those design matrices should “work”, they’re not ideal; they’re what’s referred to as rank-deficient. In both cases, you can see that the first column can be constructed as a linear combination of the second and third columns. It is therefore impossible for the linear regression to estimate un…

Great. I gotcha. Thanks for your help @rsmith and @diagiraldo As I understand it from http://mrtrix.readthedocs.io/en/latest/fixel_based_analysis/ss_fibre_density_cross-section.html It still needs a first column of 1s added - unlike FSL which does it automatically for you. So in both the cases ab…

I found that particular wording RE: “correlation analysis” unhelpful myself when I saw it in the code & documentation. The key point is that unlike FSL, a column of ones will not be automatically added in all circumstances: it is left up to the creator of the design & contrast matrices whether or no…

Thanks @rsmith That’s perfectly clear now. I did get confused by it and couldn’t make out the correct interpretation from the code. Cheers again.

Hi all, I’m stuck at step 21: fba-link for computing statistical analysis. If I have control1.mif, patient1.mif, control2.mif, patient2.mif … : Is it right to change the filenames in the fd folder to c1.mif,c2.mif,p1.mif,p2.mif and make a design_matrix.txt with ‘1 0’ ‘1 0’ ‘0 1’ ‘0 1’… ? and the…

Hi Lucius, In your example, you have: c1.mif 1 0 c2.mif 1 0 p1.mif 0 1 p2.mif 0 1 ------------------ Contrast 1 -1 Your hypothesis is therefore that the mean of your metric of interest is larger in the group consisting of [c1.mif, p1.mif] than in the group con…

Dear Rob, thank you very much for your answers, the papers and for the info concerning the order. I’m still a bit confused though: I’d like a hypothesis for two groups: c, control and p, patients. Therefore I’d like to compare the c.mif's with the p.mif's In your answer there’s a group consistin…

MRtrix3 Community

Statistical analysis of FD, FC, and FDC

show post in topic

Home
Categories
Guidelines
Terms of Service
Privacy Policy

Powered by Discourse, best viewed with JavaScript enabled