 |
 |
Hi Angelika You'd like to correlate the FA at each voxel with the MD at that same voxel, across subjects (or, instead of FA and MD, other quantities, and/or deviations from a voxelwise database). For this, I suggest using the options in randomise for voxelwise regressors, i.e., --vxf and --vxl. All the best, Anderson On 7 May 2014 14:49, Mennecke, Angelika <[log in to unmask]>wrote: > Hi Anderson, > > > > regardless of the projection problem (to which I probably come back later) > - how would you advise to calculate the correlation between the deviation > of FA from norm database and the deviation of MD from norm database? Using > the raw tbss_tstat or the corrp - images? > > > > All the best and thanks a lot, > > > > Angelika > > > > *Von:* FSL - FMRIB's Software Library [mailto:[log in to unmask]] *Im > Auftrag von *Anderson M. Winkler > *Gesendet:* Mittwoch, 7. Mai 2014 15:21 > *An:* [log in to unmask] > *Betreff:* Re: [FSL] AW: [FSL] AW: Re: AW: [FSL] Single Patient TBSS > analysis > > > > Hi Angelika, > > Hmm, I'm not sure if I understand... it's probably not possible to project > accurately thickness to the skeleton (and probably not meaningful). Maybe > if you give more details... > > All the best, > Anderson > > > > On 7 May 2014 13:11, Mennecke, Angelika <[log in to unmask]> > wrote: > > Hi Anderson, > > > > that actually was the point: the "false" positives from the first run > really disappear as I control very stringently for equality in sequence > parameters. > > > > Another question: I want to correlate the deviation of the FA from the > norm database with other parameters (deviation of GM- thickness/volumetric > parameters from normdatabase), projected on the skeleton. Should I > therefore use the raw tbss_tstat -files or the corrp as well? > > > > Thanks! > > > > Angelika > > > > *Von:* FSL - FMRIB's Software Library [mailto:[log in to unmask]] *Im > Auftrag von *Anderson M. Winkler > *Gesendet:* Mittwoch, 7. Mai 2014 13:00 > *An:* [log in to unmask] > *Betreff:* Re: [FSL] AW: Re: AW: [FSL] Single Patient TBSS analysis > > > > Hi Angelika, > > Hi Anderson and Mark, > > > > thank you very much for your valuable responses. > > > > Since I want to focus on local effects it seems quite reasonable to me to > prefer FA mean of the skeleton over the FA mean of the whole brain as a > nuisance regressor. Is this correct? > > > Yes, that's the same I'd use. > > As I do so, the rate of false positives seems to increase as well. How > can I control for this? > > > How do you know that some or all of the positives are false? Using the > corrp files at 0.05 (i.e. 0.95), the chance of having a false positive > anywhere in the image is 1 in 20. To control more stringently, maybe use > 0.01, but definitely, use only use the corrp files. > > Also, make sure from the very beginning (even before starting the > analysis) that all subjects in the sample are as similar as possible, > except for what is being studied (e.g., the diagnosis). > > All the best, > > Anderson > > >