NIH study of normal brain development. Using the general linear model, we statistically isolate the influence of external variables of interest on brain structure, allowing us to generate high-quality matched templates for any given group of subjects. The toolbox offers two options:The TOM toolbox takes a radically new approach towards providing reference data, based on imaging data from the
- to create pediatric templates (T1) and tissue maps (GM, WM, and CSF) based on the objective 1 NIH data (n = 404), in the age range of 5-18 years, or
- to assess a new reference population with regard to your variables of interest.
Of note, this approach is generally applicable and in no way restricted to analyzing pediatric imaging data: for example, if you aim at investigating the effects of aging in elderly subjects, the toolbox will also allow you to create more appropriate reference (if your group is large enough to isolate such effects).
Template creation method
Two general approaches seem feasible to construct appropriate reference data. First, the average age, gender, etc. is calculated based on the supplied input information (i.e., the demographic variables of the sample under study), and a fitting average template is created accordingly. Here, we term this the average approach. Alternatively, the input sample could be completely matched such that one reference tissue map is generated for each input subject, and these matched reference maps would only be averaged at the end. We term this the matched pairs approach.
For all template files we also save the according mat-file. Although, this is non-standard for nifti-images, it will offer backwards compatibility to older SPM versions. SPM5 (and all other nifti-based software) will simply ignore the mat-file.
Order of polynomial regression
Age can be modeled as polynomial regression with up to third order terms. The simplest model is a linear regression (which is not recommended). Either third or second order regression is appropriate for modeling aging effects. The different age terms will be orthogonalized with regard to its
This toolbox is the result of a joint effort by the Department of Pediatric Neurology and Developmental Medicine (Marko Wilke, Tuebingen, Germany), the Imaging Research Center (Scott Holland and Mekibib Altaye, Cincinnati, OH, USA), and the Structural Brain Mapping Group (Christian Gaser, Jena, Germany). The rationale, approach and further details are available in a recent NeuroImage paper.
We make the software available at the Imaging Research Center’s website. We will ask you to complete a registration form before downloading.
- unpack the zip-file
- remove the old TOM folder
- copy or link the TOM folder to the spm8/toolbox directory
You can start the toolbox by choosing “TOM” in the “Toolbox” selector.