WP 03 – Translational Research Development
- Translational neuroimaging: Development of reliable and valid imaging paradigms that can be applied across species to assess genotyped individuals, high-risk subjects and patients with ASD for functional and neuroanatomical deficits (e.g. ‘connectivity’) and circuits linked to core abnormalities of ASD, using MRI, fMRI and electrophysiological techniques. Development of human cognitive and behavioral neuroimaging paradigms for use in WP4. CNV Approaches will be a major biological axis to tie different approaches across specie. Hence we will assess a cohort of genotyped individuals, both humans (mainly from the Icelandic sample provided by deCODE) and rodents carrying the same mutations, e.g. CNVs, for functional and neuroanatomical deficits (e.g. ‘connectivity’) using resting fMRI and DT-MRI, sometimes together with electrophysiological techniques (including ERP, within scanner EEG, PPI, and CEB) used (wherever possible) in WP 4. In human patients and high-risk allele carriers, the cognitive battery will additionally be deployed and brain function measured using the same EEG, imaging, and eye-tracking paradigms as WP 4.
- Molecular neuroimaging: Development of PET methodology to assess glutamate, GABA and serotonergic function, combined with MRS measures of absolute glutamate and GABA receptor binding and release, as well as potentially novel ligands in the neuropeptide system levels, and link these molecular markers to genetic risk and disease expression in ASD. To assess the potential of TSPO ligands as indicators of neuroinflammation in ASD.
- Biomarker integration: To develop multivariate statistical methods to construct, identify and validate multimodal biomarkers that integrate imaging, genetic and proteomic signatures and link to phenotypic dimensions for segmentation/stratification of patient groups and novel compounds.
Academic Lead: Professor Dr. Andreas Meyer-Lindenberg, Central Institute of Mental Health Mannheim, Germany
EFPIA Lead: Dr. Gahan Pandina, Janssen Pharmaceutica, USA
- Central Institute of Mental Health Mannheim
- Janssen Pharmaceutica
- F. Hoffmann- La Roche
- King's College London
- Cambridge University
- deCode Genetics
- Karolinska Institutet
- Pfizer Limited
- Extend the use of categorical (Support Vector Machine) and probabilistic (Gaussian Process Classification (GPC)) methods from our previous work to this project. This will yield multivariate maps of areas driving classification as well as sensitivity and specificity data on classifications and individual categorical and probabilistic predictions.
- Apply regression (Gaussian Process Regression) as well as classification methods where continuous-valued, rather than categorical data are available.
- Use newly available techniques such as Sparse Multinomial Logistic Regression to facilitate the production of more informative multivariate maps than have been possible using pre-existing methods such as SVM and GPC. This is effectively a multivariate map thresholding method.
- Use multi-kernel methods to integrate data across modalities (MRI, PET, EEG, genetic, behavioral data). (see “WP03 Synergies between Academia and Industries”). As for timelines, we may want to group this together with the genetic assessments under “Human imaging genetics data analysis concluded