MRI template for children

Joan_Paul_pozuelos · October 2, 2014, 12:19pm

Hello everyone,

Im wondering if someone has childrens MRI template to do source localization?. Im currently working with the GSN 128 for children betwen 4 and 7 years old and havent been able to found a template that can be used in brainstorm.

I will appreciate every help you can provide

Cheers,

Joan Paul Pozuelos

Francois · October 2, 2014, 1:44pm

Hello,

I will integrate soon the following baby atlas to Brainstorm, but I'm not sure it would match your age target:
http://www.ncbi.nlm.nih.gov/pubmed/24862070
http://www.unicog.org/pm/pmwiki.php/Site/InfantTemplate

For older children, you may try to contact the authors of the following articles:
http://www.jove.com/video/51705/cortical-source-analysis-of-high-density-eeg-recordings-in-children
http://www.sciencedirect.com/science/article/pii/S1053811913006381

Cheers,
Francois

Joan_Paul_pozuelos · October 3, 2014, 12:54am

Thanks for the quick response Francois :)…

I have found some web pages in which there are some templates. I dont know if they can be uploaded also into brainstorm. The webpage is:
http://www.bic.mni.mcgill.ca/ServicesAtlases/NIHPD-obj1

Do you recommend to use this templates?. im very new in using braintstorm so i dont really have experience in uploading a template.

Thank you for your help!

Cheers,
Joan Paul

Francois · October 3, 2014, 8:17am

Hi Joan Paul,

The problem with those atlases is that they are all transformed to the MNI stereotaxic space.
If you process them in FreeSurfer and import them in Brainstorm, they will all have the same size and almost the same shape.

What you need for source reconstruction is instead an anatomy that is the real size of the kid’s head. If not, you will estimate brain activity outside of the head.
An alternative is to take an existing template, and deform it to fit the size and shape of the children’s heads. For that, you just need to digitize the envelope of the head with a Polhemus (or equivalent tracking device). This would not give you an accurate folding pattern or BEM model, but at least it will give you a head model of the appropriate size. Use only unconstrained source models if you do this.
See these tutorials:
http://neuroimage.usc.edu/brainstorm/Tutorials/TutDigitize
http://neuroimage.usc.edu/brainstorm/Tutorials/TutWarping

Cheers,
Francois

giovanni.mento · October 6, 2014, 8:46am

Dear Francois,
I would be too interested in pediatric cortical sources estimation (6 to 12 year-old-children). I found this public database: http://jerlab.psych.sc.edu/NeurodevelopmentalMRIDatabase/index.html

Are there the same problems as those you raised for the McGill template?How , if possible, could I implement those templates on BS?
Alternatively, are you planning to integrate children templates in BS?
Is it completely wrong to use adult templates for children data?

Thank you very much

Giovanni

Francois · October 7, 2014, 8:53am

Hi Giovanni,

You should contact the authors and ask them if the average MRI is in a normalized space or if it is in real size.
If they are not normalized to the MNI space, you can process them with FreeSurfer and import then in Brainstorm as you would do for any individual MRI:
http://neuroimage.usc.edu/brainstorm/Tutorials/LabelFreeSurfer

If you use adult-size heads for studying children, you will reconstruct sources up to 5cm outside the subject’s brain, then the results can be quite random…
You should at least scale/warp one of the templates using head points, as explained in these tutorials:
http://neuroimage.usc.edu/brainstorm/Tutorials/TutWarping
http://neuroimage.usc.edu/brainstorm/Tutorials/TutDigitize

Francois

giovanni.mento · October 7, 2014, 8:57am

thank you very much Francois,
I’ll ask the author and then I will tell you!

Giovanni

ps:anyway, are you planning to implement children templates in BS?

Francois · October 7, 2014, 9:02am

The development plans include the baby atlas developed at Neurospin:

ncbi.nlm.nih.gov

Anatomical correlations of the international 10-20 sensor placement system in infants.

C Kabdebon, F Leroy, H Simmonet, M Perrot, J Dubois and G Dehaene-Lambertz, NeuroImage, Oct 2014 01

Developmental research, as well as paediatric clinical activity crucially depends on non-invasive and painless brain recording techniques, such as electroencephalography (EEG), and near infrared spectroscopy (NIRS). However, both of these techniques measure cortical activity from the scalp without precise knowledge of the recorded cerebral structures. An accurate and reliable mapping between external anatomical landmarks and internal cerebral structures is therefore fundamental to localise brain sources in a non-invasive way. Here, using MRI, we examined the relations between the 10-20 sensor placement system and cerebral structures in 16 infants (3-17 weeks post-term). We provided an infant template parcelled in 94 regions on which we reported the variability of sensors locations, concurrently with the anatomical variability of six main cortical sulci (superior and inferior frontal sulcus, central sulcus, sylvian fissure, superior temporal sulcus, and intraparietal sulcus) and of the distances between the sensors and important cortical landmarks across these infants. The main difference between infants and adults was observed for the channels O1-O2, T5-T6, which projected over lower structures than in adults. We did not find any asymmetry in the distances between the scalp and the brain envelope. However, because of the Yakovlean torque pushing dorsally and frontally the right sylvian fissure, P3-P4 were not at the same distance from the posterior end of this structure. This study should help to refine hypotheses on functional cognitive development by providing an accurate description of the localization of standardised channels relative to infants' brain structures. Template and atlas are publicly available on our Web site (http://www.unicog.org/pm/pmwiki.php/Site/InfantTemplate).

http://www.unicog.org/pm/pmwiki.php/Site/InfantTemplate

I didn’t have any proposition for adding other children templates yet.