Overlap between metabolic and other measures of brain connectivity: a multimodal imaging study

Background: Metabolic connectivity (MC) aims to detect functionally interacting brain regions based on PET recordings with the glucose analogue [F]fluorodeoxyglucose (FDG). While the method becomes increasingly popular, relationship of MC with other measures of brain connectivity remains unclear. The aim of this study was to examine a degree of overlap of MC with functional, structural connectivity, as well as cortical thickness’ covariance.

Methods: Structural MRI, diffusion tensor imaging, functional MRI (fMRI) and FDG-PET data of 200 subjects were acquired on a hybrid PET/MR system. PET data were corrected for partial volume effects. Resting state frequencies were extracted from fMRI data after correcting for motion, slice-timing, and filtering by a bandpass time filter. Volumetric-based cortical thickness was estimated using Advanced Normalization Tools (ANTs). All images were spatially normalized to a standard space and segmented into 62 anatomical regions using a probabilistic atlas. Sparse inverse covariance estimation was employed to quantify metabolic connectivity. Structural connectivity was measured by streamline tractography through fiber assignment by continuous tracking. Functional connectivity was estimated on the mean of Pearson correlation of all subjects. The covariance in cortical thickness was calculated using the Pearson correlation as well. Finally, the same number of connections for each type of connectivity measure (network) was generated. Results: All 4 networks had a significantly higher convergence ratio (CR) and similarity metrics compared to randomly generated networks. CR was defined as a ratio of convergent connections in any combination of networks (modalities) to all detected connections (in any network). Above 85 % of all possible connections were similar, i.e. either present or absent in any pair of networks. CR was around 60 %, 50 %, and 40 % for cortical thickness’ covariance, structural, and functional connectivity, respectively. An overlap between all 4 measures of brain connectivity was present in only 16 % of all possible connections. Conlusions: Our preliminary findings indicate that contrary to a common view MC may be closer to the structural rather than functional domain of brain connectivity. Furthermore, there is only a limited overlap between the different measures of brain connectivity, indirectly pointing at their unique value. These results have important implications for understanding brain function. Further analyses are ongoing.