Title: Optimizing preprocessing pipelines in PET/MRI neuroimaging
PhD-student: Martin Nørgaard, NRU
Neuroimaging represents a powerful and important clinical tool that can generate novel information about the neurobiology and diseases of the brain. There is now, however, a growing concern that neuroimaging outcomes often are surprisingly difficult to replicate and that the choice of preprocessing steps (”the pipeline”) has a major impact on the outcome and conclusions drawn. We have in preliminary recent research demonstrated that optimizing the neuroimaging data analysis pipeline for fMRI significantly increases sensitivity and reliability when detecting inter-subject and group differences.
We will extend this previous work to include multiple modalities by using an already acquired unique data set, including brain-PET, structural MRI, and relevant behavioural measures for 110 healthy controls that all have been scanned twice (3, 5, or 26 weeks apart). The effect of preprocessing choices on signal detection and reliability will be optimized by quantitatively evaluating processing pipelines and their complex component interactions. The improved and stringent processing pipeline can subsequently be addressed to pertinent issues in the healthy brain as well as to disclose subtle changes between healthy subjects and patients with brain disorders.
The aim of the project is thus to augment the power of preprocessing choices in clinical neuroimaging and we strongly believe that the proposed work, even in the presence of relatively small sample sizes, will lead to more reproducible results and thereby enable researchers to draw correct conclusions.