Dear Jeff,
On 17 Aug 2018, at 17:02, Jeff Browndyke wrote:
> Has anyone here used SwE on CAT12-derived data, particularly
> longitudinal segmentations and tissue maps?
>
> If so, any recommendations on how to implement SwE with data from
> Christian’s segmentation and surface analysis toolbox?
There should be nothing special with data from CAT12. The SWE toolbox
seems to be also prepared for meshes, thus you can also analyze surface
data.
Best,
Christian
>
> Thanks to both Tom and Christian for creating and sharing these
> toolboxes with the neuroimaging community!
>
> Warm regards to all,
> Jeff
>
> ——————
> Jeff Browndyke, Ph.D.
> Duke University Medical Center
>
>
>> On Aug 17, 2018, at 4:40 AM, Thomas Nichols
>> <[log in to unmask]> wrote:
>>
>> Hi folks,
>>
>> The SwE Toolbox for SPM - http://www.nisox.org/Software/SwE
>> <http://www.nisox.org/Software/SwE> - an implementation of the
>> "marginal model" approach to repeated measures and longitudinal data,
>> has a new update, version v1.2.9. Download this latest version at:
>> http://www.nisox.org/Software/SwE/download
>> <http://www.nisox.org/Software/SwE/download> .
>>
>> While the models/inferences are unchanged, there are significant
>> improvements to the user interface:
>> The wild bootstrap, used to obtain cluster inferences, now has a
>> SPM-style results page (before results were only available as NIFTI
>> images, with no results page).
>> Results page has informative footer, summarising the data and design,
>> e.g. listing number of subjects, visits, etc.
>> File name conventions changed, now consistent and clearly documented
>> in the manual <http://www.nisox.org/Software/SwE/man>.
>> When running the wild bootstrap, a large number of "fit" and "resid"
>> images were left around; these are now removed.
>> There are numerous other changes to code structure that will make
>> easier for us to maintain the work going forward. Full details are in
>> the release notes
>> <https://github.com/NISOx-BDI/SwE-toolbox/releases/tag/v1.2.9>.
>>
>> Please contact us with any questions via the SwE-Toolbox Support
>> <https://groups.google.com/forum/#!forum/swe-toolbox> group email.
>>
>> -Tom Nichols, Tom Maullin-Sapey & Bryan Guillaume
>>
>>
>> More detail on SwE: SPM can accommodate repeated measures and
>> longitudinal data but with a stringent assumption that the
>> intra-subject correlation is the same over the whole brain. Depending
>> on the model, SPM will also fit per-subject intercepts (dummy
>> variables), precluding the use of any between subject variables --
>> e.g. gender or age. In contrast, SwE uses no per-subject intercepts,
>> and any collection of within- or between-subject variables desired
>> can be used. Ordinary least squares is used to estimate the
>> regression model and the repeated measures/longitudinal dependence is
>> estimated and used to correct the standard errors of the regression
>> estimates. The only assumption that the SwE uses is that dependence
>> among repeated measures/visits is the same for all subjects in a
>> group (SwE type "modified", the default), though if enough subjects
>> (100+) are available, then absolutely no assumptions on the
>> dependence need to be made (select SwE Type "classic"). The result is
>> not as statistically efficient as full-blown linear mixed effects
>> models (e.g. as in R's lmer) but in our evaluations [Guillaume et al.
>> 2014] we found SwE and LME to be virtually indistinguishable -- plus,
>> LME rarely considers any dependence structure more complex than
>> random intercept and slope, while SwE accommodates any repeated
>> measures dependency structure. While Random Field Theory results are
>> not available for this class of models, voxel-wise FDR corrected
>> inferences are provided and, with the nonparametric wild bootstrap,
>> FWE-corrected voxel- and cluster-wise inferences are available.
>> Guillaume, B., Hua, X., Thompson, P. M., Waldorp, L., Nichols, T. E.,
>> & Alzheimer’s Disease Neuroimaging Initiative. (2014). Fast and
>> accurate modelling of longitudinal and repeated measures neuroimaging
>> data. NeuroImage, 94, 287–302.
>> http://doi.org/10.1016/j.neuroimage.2014.03.029
>> <http://doi.org/10.1016/j.neuroimage.2014.03.029>
>> __________________________________________________________
>> Thomas Nichols, PhD
>> Professor of Neuroimaging Statistics
>> Nuffield Department of Population Health | University of Oxford
>> Big Data Institute | Li Ka Shing Centre for Health Information and
>> Discovery
>> Old Road Campus | Headington | Oxford | OX3 7LF | United Kingdom
>> T: +44 1865 743590 | E: [log in to unmask]
>> <mailto:[log in to unmask]>
>> W: http://nisox.org <http://nisox.org/> | http://www.bdi.ox.ac.uk
>> <http://www.bdi.ox.ac.uk/>
|