 |
 |
Hi Anderson, Thanks for you answer. I'd like to be a bit more precise. If I look at the uncorrected p-value for the TFCE results for FA threshold = 0.2 and FA threshold =0.01, I can see good overlap in the region where differences are significant. However after FWE correction, some regions for FA threshold = 0.2 become not significant while they are in the case of FA threshold = 0.01 and vice versa. As you said, the TFCE results depends on the support region and so I don't expect for different FA threshold to get the same results but if I have very similar uncorrected p-values, I was wondering how was FWE and size of interested regions are linked. Hope that is bit more clear! Thank you so much! Giang-Chau 2016-09-29 3:16 GMT-05:00 Anderson M. Winkler <[log in to unmask]>: > Hi Giang-Chau, > > The TFCE results are already FWE corrected. Unless you looked into the > uncorrected, something should not be done. > > All the best, > > Anderson > > > On 29 September 2016 at 03:21, Giang-Chau Ngo <[log in to unmask]> wrote: > >> Hi Anderson, >> >> Thank you for your quick answer! I understand now. >> Do you think the FWE correction could lead to some differences too since >> the number of voxels included are different? >> >> Thanks! >> >> All the best, >> >> Giang-Chau >> >> 2016-09-28 1:51 GMT-05:00 Anderson M. Winkler <[log in to unmask]>: >> >>> Hi Giang-Chau, >>> >>> The support region of the TFCE (see the original paper >>> <http://dx.doi.org/10.1016/j.neuroimage.2008.03.061>) depends on the >>> size of the region one is looking at, which here depends on the FA >>> threshold, so the results can differ. >>> >>> All the best, >>> >>> Anderson >>> >>> >>> On 27 September 2016 at 21:52, Giang-Chau <[log in to unmask]> wrote: >>> >>>> Dear FSL experts, >>>> >>>> I have two infants DTI data set and I have run TBSS + randomise (with >>>> TFCE) analysis using different FA thresholds (0.1, 0.15, 0.2) therefore >>>> different masks. >>>> When looking at the corrected p-value map, I have noticed that some >>>> tracts appear significantly (p<0.05) different when using a FA=0.02 but are >>>> not when using=0.1 and vice versa. Are the differences in p-value between >>>> the different thresholds resulting from the correction of the multiple >>>> comparison? >>>> >>>> Thank you very much for your time and help! >>>> >>>> Giang-Chau >>>> >>> >>> >> >