Combining structural and functional MRI data: biases and solutions when comparing structurally different cohorts

• Applicant: Professor Dr. Marko Wilke
• Subject Area: Pediatric and Adolescent Medicine
• Term: from 2010 to 2014
• Project identifier: Deutsche Forschungsgemeinschaft (DFG) - Project number 157786980

Comparing individuals using functional MR imaging requires a global and (usually) regional matching of the MRI data in size and orientation. this process is called spatial normalization. We could previously show that using inappropriate (adult) reference data during this process results in a severe bias when normalizing pediatric MR imaging data. In the previous funding period, we could recruit more than 80 children aged 8-17 years, either born very preterm or healthy (term-born) control children. For all children, extensive clinical and neuropsychological information is available, and both structural and functional MRI data has already been obtained. Preliminary analyses now strongly suggest that functional group differences between the two cohorts are either masked or exaggerated as a result of underlying structural group differences. This project is now aimed at developing an algorithm that allows to identify and visualize this potentially severe bias for functional MRI group analyses. Further, a Baysesian approach towards correcting this bias is suggested.

• DFG Programme: Research Grants