Bimodal and multivariate analysis linking FDG uptake to BOLD signal at rest: a simultaneous hybrid PET/MR study

Abstract

Glucose consumption and hemodynamic response to neural activity at rest reflect intrinsic brain function in general. Investigating the relationship between them should be needed to better understand how the brain works. I applied the multivariate partial least squares for investigating the relationship between FDG uptake and BOLD signal at rest (n = 38, mean age 44 ± 13.9 years), as representative measurements for capturing those different aspects of brain function using a hybrid PET/MR scanner. This approach can consider the interdependency between voxels as well as the relationship between modalities. As a result, the relationship which was reproducible by a split-half resampling, was found in brain areas in respect to their functional roles. For example, the FDG uptake and BOLD signal showed positive relationship within the brain regions such as (1) sensory system, and (2) default mode network system. The negative relationship was found between the above regions. The principal component analysis was also performed to investigate the interregional correlation patterns within modality. The first principal component image of FDG-PET was almost the same to the first singular image of FDG-PET. While, the first and second principal component images of fMRI were similar to the first singular image of fMRI. The first principal component image of fMRI represented the (1) visual system, and (2) default mode network system. The second principal component image of fMRI represented the (1) auditory and motor systems, and (2) default mode network system. The results proved that the interregional correlation patterns derived by the spontaneous ongoing neural activity were divided into the specialized independent systems compared to the interregional correlation patterns derived by the relatively stable FDG uptake. This present work showed that the relationship between FDG uptake and BOLD signal was changed by age. By applying the partial least squares to the younger adults (n = 19, mean age 32 ± 6.9 years) and older adults groups (n = 19, mean age 56 ± 7.7 years) respectively, the different relationship between FDG uptake and BOLD signal was observed. The changes of brain function due to aging can be observed by the changes of the relationship between FDG uptake and BOLD signal.