Work Package 6
Methodological aspects of longitudinal imaging genetics MRI studies: reliability, quality assurance, statistical genetics and genetic epidemiology
Prof. Dr. Andreas Jansen1, Prof. Dr. Astrid Dempfle2 1University of Marburg, Department of Psychiatry and Psychotherapy
2University of Marburg, Institute of Medical Biometry and Epidemiology
WP6 has two major aims. The first aim is the implementation of a comprehensive quality assessment (QA) protocol for MRI data. The second aim is the development of statistical analysis strategies for the FOR. WP6 will further act as biostatistical and methodological platform for all other WPs. Part 1: Quality assessment protocol: Although modern MRI systems show good technical quality, image characteristics can change significantly over the course of a study. This is in particular a major challenge for functional magnetic resonance imaging (fMRI) studies since functional signal changes are typically just a small fraction (~1-5 %) of the raw signal intensity. In the present project it is planned to measure a large cohort of patients and control subjects with functional and structural MRI in a longitudinal study design. Stable scanner performance is therefore required over years to differentiate for instance between MRI signal changes that are associated with the time course of a disease and signal changes caused by alterations in the MR scanner environment. The aim of the first part of WP6 is therefore the systematic implementation of a comprehensive QA protocol for MRI data that monitors scanner performance, defines scanner benchmark characteristics, documents changes in scanner hardware and software, and serves as an early-warning system for potential scanner malfunctions. Major aspects of this protocol, apart from the regular measurement of MRI phantoms, will be the automation of the data analysis, the public documentation of the MRI QA data and the regular monitoring of the adherence to the protocol. Part 2 will provide the methodological tools and expertise to fully exploit the wealth of complex data generated within this FOR through state-of-the-art statistical methods. We have three main goals: first, identification of genetic variants associated with structural and functional MRI variations (in particular amygdala activation and hippocampal morphometry) in patients with affective disorders (AD)(recruited and phenotyped in WP1) and the modification of their effects by environmental exposures. A major methodological challenge of this task is the high dimensionality of data both on the phenotypic (MRI) and on the genotypic level, where genome-wide marker data will be provided by WP5. New statistical analysis approaches to account for this will be developed and implemented. The second goal is the definition of biotypes, i.e. new etiologic entities within AD beyond the clinical disease definitions of DSM and ICD. Statistical analysis strategies for clustering multi-dimensional biological, cognitive, psychopathological longitudinal data will be developed. Third, WP6 will provide statistical support for all other WPs, e.g. for the comparison of microRNA profiles (WP3), histone modification (CP1) and gene-environment interactions of complex behavioural phenotypes in rats (WP2).