CIBC:Workshops:Workshop06:summ-weber

From NCRR Biomedical Software Development, Engineering, and Dissemination Wiki

Summary for Darren Weber

Affiliation

Darren L. Weber, Ph.D. Postdoctoral Scholar

Dynamic Neuroimaging Laboratory,
UCSF Department of Radiology,
185 Berry Street, Suite 350, Box 0946,
San Francisco, CA 94107, USA.

Tel: +1 415 353-9444
Fax: +1 415 353-9421
www: http://dnl.ucsf.edu/users/dweber

Summary

I currently hold a postdoctoral position at UCSF. We are working with a CTF EEG/MEG system that can acquire 275 channels of MEG (radial gradiometers) and 128 channels of EEG. The signal processing demands of this data are significant. We struggle to process this data in common matlab packages, such as EEGLAB, at high temporal resolution, simply due to memory limitations. I would like to run ICA and time-frequency analyses on these data, using c/c++ tools that will compile for linux and sun systems. In addition, I have been working on integration of these data with anatomical models derived from high resolution MRI, following the work of Dale & Sereno (1993) for distributed electromagnetic source estimation. These integration approaches are computationally intensive. Using the freesurfer package from the MGH, it can take several days to process an anatomical dataset simply to extract a detailed model of the cortical surface. For boundary element modeling (BEM) of source activity on the coritical surface, we also require good anatomical models of the skull and scalp surfaces (for EEG). There are some efforts to obtain these surfaces (eg, the FSL BET2 program). The calculation of the MEG/EEG forward solution (ie, calculating sensor data for any given source configuration) is very intensive for BEM and we could benefit from solid routines to generate these data (I am currently using the BrainStorm toolbox for matlab). In addition, the visualization of cortical temporal-spatial dynamics is challenging, because details of visualization are not trivial. The cortical surface is convoluted, so it is often inflated slightly to provide views of the sulcal areas. More importantly, the choice of scale, thresholds and coloring of cortical activity is very important. It is too easy to distort valuable data with poor visualizations. Finally, the results of brain source estimation need to be evaluated for statistical effects, within and across subjects. In this regard, tools to implement the permutation exact tests of Blair & Karnisky (1993 and related papers) would be valuable. I would like to learn more about how the Utah developments have solved these problems and this workshop should provide a short intensive program that can introduce me to the packages provided. I did not get to a previous workshop a few years ago.