Trimodal imaging of human brain networks using simultaneous PET/MR/EEG
Final Report Abstract
The trimodal study of hd-EEG/fMRI/[18F]FDG-PET was analyzed for patients with temporal lobe epilepsy. A series of preprocessing steps are taken before data analysis, and fMRI and [ 18F]FDG-PET modalities show similar activation patterns. However, when separating the data into two cohorts (patients and controls), patient activation was significantly lower. Hypometabolism masks are created at different thresholds and show heterogeneity, concluding that the cohorts need to be larger and within the same age range to create a homogeneous mask. The connectivity analysis of the study shows connectivity networks only for controls in key regions of interest. Further analysis is required to find effective connectivity by correlating [18F]FDG-PET maps with seed-to-voxel fMRI maps. In addition, we showed that spatially coherent resting-state network can be identified from each modality of a simultaneous trimodal hd-EEG/fMRI/[18F]FDG-PET measurement. Most of vascular fMRI networks, were also found in neuronal hd-EEG data, reinforcing the hypothesized link between vascular signaling and neuronal activity. Each modalities measurement property (metabolic, vascular, neuronal), as well as their spatial and temporal resolution are crucial variables that influence the identification of the RSN in the brain in terms of their spatial profile and the occurrence of networks. Nevertheless, the fact of cross-modal expression of some RSN reinforces further investigation of such analyses of simultaneously measured trimodal (hd-EEG/fMRI/[18F]FDG-PET) datasets, as they hold the potential to identify biomarkers for neuronal disease in multiple spatial and temporal dimensions. Overall, we conclude that the project aims of providing a better understanding of resting-state and taskrelated network of the human brain by a trimodal study is feasible and can yield new insights into brain functions. However, such work is logistically challenging and had to overcome several hurdles. We are currently working on finalizing the two sub-projects as manuscript to be submitted for peer-review. Furthermore, we are working on making this unique dataset available for interested partners for re-use according to FAIR data sharing principle.
