Imaging transcriptomics: Convergent cellular, transcriptomic, and molecular neuroimaging signatures in the healthy adult human brain

The expansion of neuroimaging techniques over the last decades has opened a wide range of new possibilities to characterize brain dysfunction in several neurological and psychiatric disorders. However, the lack of specificity of most of these techniques, such as magnetic resonance imaging (MRI)-derived measures, to the underlying molecular and cellular properties of the brain tissue poses limitations to the amount of information one can extract to inform precise models of brain disease. The integration of transcriptomic and neuroimaging data, known as imaging transcriptomics, has recently emerged as an indirect way forward to test and/or generate hypotheses about potential cellular and transcriptomic pathways that might underly specific changes in neuroimaging MRI biomarkers. However, the validity of this approach is yet to be examined in-depth. Here, we sought to bridge this gap by performing imaging transcriptomic analyses of the regional distribution of well-known molecular markers, assessed by positron emission tomography (PET), in the healthy human brain. We focused on tracers spanning different elements of the biology of the brain, including neuroreceptors, synaptic proteins, metabolism, and glia. Using transcriptome-wide data from the Allen Brain Atlas, we applied partial least square regression to rank genes according to their level of spatial alignment with the regional distribution of these neuroimaging markers in the brain. Then, we performed gene set enrichment analyses to explore the enrichment for specific biological and cell-type pathways among the genes most strongly associated with each neuroimaging marker. Overall, our findings show that imaging transcriptomics can recover plausible transcriptomic and cellular correlates of the regional distribution of benchmark molecular imaging markers, independently of the type of parcellation used to map gene expression and neuroimaging data. Our data support the plausibility and robustness of imaging transcriptomics as an indirect approach for bridging gene expression, cells and macroscopical neuroimaging and improving our understanding of the biological pathways underlying regional variability in neuroimaging features.