 |
 |
Thanks, Mark for your helpful reply on the demeaning -D option. After running the TBSS, I am interested in extracting the FA voxels in the particular region in the brain which showed significant difference between the pre-term and term control infants; and then plot those FA voxels data against the cognitive scores. Is there a way to extract out the sigificant voxels and then do the statistics. I'd appreciate your help. Thanks, Supreet On Mon, Aug 12, 2013 at 3:40 AM, Mark Jenkinson < [log in to unmask]> wrote: > Hi, > > If all you are interested in is the relation with the covariate and not > the overall mean then you can use the -D option in randomise, but if you > are doing anything more complicated then you will have to remove the mean > from the covariate values yourself (i.e. calculate the mean value and > subtract it, by hand, or with SPSS, or matlab, or octave, or excel, …) > > All the best, > Mark > > > On 11 Aug 2013, at 20:46, Supreet kaur <[log in to unmask]> wrote: > > Dear FSL experts, > > I am trying run the TBSS - randomize for two group with continuous > covariate interaction, is there any option with the GLM gui > > that could demean the covariate values? > > Thanks i advance, > Supreet > > > On Sun, Aug 11, 2013 at 3:23 PM, Supreet kaur <[log in to unmask]> wrote: > >> Hi Mark, >> >> If I need to extract FA values from a ROI mask which I created for a >> particular region inside the brain, the left >> Wernick's area, how is it possible? should I be using the fslroi or >> fslmeants command? >> and how do I get the statistics for the FA values? >> >> Thanks, >> Supreet >> >> >> On Sat, Aug 10, 2013 at 4:37 AM, Mark Jenkinson < >> [log in to unmask]> wrote: >> >>> Hi, >>> >>> You can use fslmerge to make a 4D image from a set of 3D ones. >>> You can extra FA values from any voxel or ROI using fslmeants. >>> >>> All the best, >>> Mark >>> >>> >>> On 9 Aug 2013, at 23:47, Supreet <[log in to unmask]> >>> wrote: >>> >>> Hi Anderson, >>> Is there a way that I can create a 4D data? Also after running >>> randomise, am I be able to >>> plot the FA values against the scores? >>> Thanks for your help >>> Supreet. >>> >>> Sent from my iPad >>> >>> On Aug 9, 2013, at 5:22 PM, "Anderson M. Winkler" < >>> [log in to unmask]> wrote: >>> >>> Hi Supreet, >>> >>> A number of issues here: >>> - Data is missing for more than half of the preterm infants; >>> - The pretermdata is supposed to be a 4D image, not a directory; >>> - Either include a column of ones or use -D, not both; >>> - If you have 41 subjects, you can use much more permutations, say, >>> 5000 or 10000; >>> - The -V option is no longer needed (although it doesn't hurt to leave. >>> >>> Hope this helps! >>> >>> All the best, >>> >>> Anderson >>> >>> >>> >>> On 9 August 2013 22:07, Supreet kaur <[log in to unmask]> wrote: >>> >>>> Hi Anderson, >>>> >>>> I used the single group average with additional co variate. However. I >>>> am getting these errors when I ran the analysis for 41 pre-term scores. >>>> Could you please take a look at the attached word doc. I'd really >>>> appreciate your help. >>>> Thanks, >>>> Supreet. >>>> >>>> >>>> On Fri, Aug 9, 2013 at 10:47 AM, Anderson M. Winkler < >>>> [log in to unmask]> wrote: >>>> >>>>> Hi Supreet, >>>>> >>>>> I think this is covered by the manual: >>>>> http://fsl.fmrib.ox.ac.uk/fsl/fslwiki/GLM >>>>> >>>>> In particular, have a look at these two examples: >>>>> - >>>>> http://fsl.fmrib.ox.ac.uk/fsl/fslwiki/GLM#Single-Group_Average_with_Additional_Covariate >>>>> - >>>>> http://fsl.fmrib.ox.ac.uk/fsl/fslwiki/GLM#Two_Groups_with_continuous_covariate_interaction >>>>> >>>>> The first you can use for the analysis of the pre-term only (no >>>>> comparison with the term infants). In this example, the second contrast >>>>> will show where the DTI measures are associated with the BSITD scores. >>>>> >>>>> The second example you can use for the comparison of the relationship >>>>> between BSITD and DTI between the groups, i.e., whether BSITD and DTI >>>>> relate differently to each other for the different groups. >>>>> >>>>> And of course, you can also use >>>>> http://fsl.fmrib.ox.ac.uk/fsl/fslwiki/GLM#Two-Group_Difference_Adjusted_for_Covariate >>>>> The first two contrasts in this example will tell if the groups >>>>> differ in their DTI measures taking BSITD into account as a nuisance (I >>>>> don't think this interests you much) and the remaining two contrasts tell >>>>> whether and where BSITD changes are associated with changes in the DTI >>>>> measures after taking into account the fact that the infant was born >>>>> pre-term or at term. >>>>> >>>>> This all assumes that the scores are in sync with the MRI (you still >>>>> didn't mention whether the MRI was before birth or after). Also, it isn't >>>>> clear whether "scores were taken at term" means that the scores were taken >>>>> at birth or at some point around 38-40 weeks even if born weeks earlier. In >>>>> any case, if you corrected the scores using a method that is valid and well >>>>> accepted, then the above examples should work fine for you. >>>>> >>>>> Hope this helps! >>>>> >>>>> All the best, >>>>> >>>>> Anderson >>>>> >>>>> >>>>> >>>>> On 9 August 2013 14:46, Supreet kaur <[log in to unmask]> wrote: >>>>> >>>>>> Thanks, Anderson for your prompt response. I have BSITD scores for >>>>>> per subject (41 pre-terms (less than 32 weeks) and 16 full-term infants (38 >>>>>> weeks), the pre-term have been age >>>>>> corrected and these scores were taken at term. I have each >>>>>> individual score for corresponding infant. First, I am planning to just >>>>>> correlate the pre-term infants scores with their FA and then MD images. >>>>>> >>>>>> Then in the next analysis, I'd like to do the group comparison >>>>>> between pre-term and term infants' DTI measures and their corresponding >>>>>> scores. Is it possible to do these analysis in randomize? >>>>>> I'd appreciate your help. >>>>>> Thanks, >>>>>> Supreet. >>>>>> >>>>>> >>>>>> >>>>>> On Fri, Aug 9, 2013 at 9:19 AM, Anderson M. Winkler < >>>>>> [log in to unmask]> wrote: >>>>>> >>>>>>> Hi Supreet, >>>>>>> >>>>>>> >>>>>>> On 9 August 2013 13:42, Supreet kaur <[log in to unmask]> wrote: >>>>>>> >>>>>>>> >>>>>>>> Dear FSL experts, >>>>>>>> >>>>>>>> I would like to perform the voxelwise cross subject statistics >>>>>>>> to assess the relationship between FA, AD, and RD and performance scores of >>>>>>>> the infants (pre-terms and term infants) BSITD-III for our DTI cohort. I >>>>>>>> have two concerns: >>>>>>>> >>>>>>> >>>>>>> I guess you scanned both groups (pre-term and term) after they >>>>>>> were born, is this correct? Or did you scan them before birth and you'd >>>>>>> like to compare with BSITD obtained after? >>>>>>> >>>>>>> >>>>>>> 1) For applying the tbss on the non-FA images (I want to input >>>>>>>> the MD images), how do I input all the subject MD images? after I am done >>>>>>>> with the tbss analysis on the FA data. Is >>>>>>>> inside the tbss, does the tbss_non_FA command will automatically >>>>>>>> read it as non-FA images? >>>>>>>> >>>>>>> >>>>>>> Have you checked this page: >>>>>>> http://fsl.fmrib.ox.ac.uk/fsl/fslwiki/TBSS/UserGuide#Using_non-FA_Images_in_TBSS >>>>>>> >>>>>>> >>>>>>> 2) For the randomize, since I'll be correlating the pre-term >>>>>>>> infant's FA and MD images to their mean BSITD-III scores,how can I >>>>>>>> incorporate the BSITD-III scores inside the model? is it going to be the >>>>>>>> paired t test or unpaired one? >>>>>>>> >>>>>>> >>>>>>> You need to give more information. How many BSITD scores do you >>>>>>> have per subject to take the mean? Do you have a matching image for each of >>>>>>> these scores for each subject, or just one image per infant? At which ages >>>>>>> were they (scores and/or images) taken? Are you going to check the >>>>>>> relationship with the MRI measures only for the pre-term or for both groups? >>>>>>> >>>>>>> Depending on what you have and what you'd like to do, the design >>>>>>> is going to be different. >>>>>>> >>>>>>> All the best, >>>>>>> >>>>>>> Anderson >>>>>>> >>>>>>> >>>>>>> >>>>>>>> >>>>>>>> I'd highly appreciate your feedback on this. >>>>>>>> >>>>>>>> >>>>>>>> >>>>>>>> Sincerely, >>>>>>>> >>>>>>>> Supreet >>>>>>>> >>>>>>> >>>>>>> >>>>>> >>>>>> >>>>>> -- >>>>>> Sincerely, >>>>>> >>>>>> Supreet kaur, >>>>>> Biomedical research engineer, >>>>>> Nationwide Childrens Hospital, >>>>>> Columbus, OH >>>>>> (614)355-3509 >>>>>> >>>>> >>>>> >>>> >>>> >>>> -- >>>> Sincerely, >>>> >>>> Supreet kaur, >>>> Biomedical research engineer, >>>> Nationwide Childrens Hospital, >>>> Columbus, OH >>>> (614)355-3509 >>>> >>> >>> >>> >> >> >> -- >> Sincerely, >> >> Supreet kaur, >> Biomedical research engineer, >> Nationwide Childrens Hospital, >> Columbus, OH >> (614)355-3509 >> > > > > -- > Sincerely, > > Supreet kaur, > Biomedical research engineer, > Nationwide Childrens Hospital, > Columbus, OH > (614)355-3509 > > > -- Sincerely, Supreet kaur, Biomedical research engineer, Nationwide Childrens Hospital, Columbus, OH (614)355-3509