Constructing the Human Brain Metabolic Connectome via Whole-Brain H⁺ MRSI: Methods, Robustness, Topology, and Biological Significance

The emergence of network science in neurobiology has significantly advanced our understanding of brain organization, uncovering the self-organizing, scale-invariant principles underlying its structural and functional connectivity. However, existing neuroimaging modalities, such as diffusion MRI and fMRI, fail to capture the biochemical processes sustaining these networks, leaving a critical gap in connectomics. Since brain function is fundamentally sustained by oxidative glucose metabolism, metabolic imaging offers a transformative opportunity to explore the biochemical underpinnings of brain network organization.
Inter-individual variation in brain network organization has been linked to behavioral differences, including general intelligence, working memory, and personality traits, and is increasingly recognized as critical to understanding neurological and psychiatric disorders. Many such disorders, often termed connectopathies, arise from disruptions in distributed brain networks rather than isolated regions. Central to these networks are highly connected hub nodes, which play critical roles in maintaining global network integrity. These nodes, characterized by high centrality, exhibit proportionally higher metabolic demands, making them vulnerable to pathologies that induce oxidative stress. Energy deprivation in these nodes can lead to global network failure, emphasizing the need for imaging methods capable of assessing metabolic activity across the brain.
In this presentation, we introduce the first human brain metabolic connectome, constructed using advanced 3D proton magnetic resonance spectroscopic imaging (MRSI). This state-of-the-art technique enables whole-brain, multi-metabolite imaging at high spatial resolution (5mm isotropic) with rapid acquisition times (<12 minutes), making it practical for large-scale and clinical studies. Using data from 68 healthy adolescents, validated on an independent cohort (N=13), we developed a robust pipeline for constructing metabolic similarity matrices (MeSiMs). These matrices demonstrate reproducibility across individuals, datasets, MRI scanners and anatomical parcellations and robustness to uncertainties in MRSI reconstruction.
Our analysis reveals that metabolic networks exhibit natural network properties, including modular organization and robust homotopic patterns, reflecting functionally integrated yet spatially distinct systems. Central to these findings is the identification of the metabolic fiber: a smooth homotopic gradient of metabolic similarity that traverses major brain regions from the occipital lobe through the parietal lobe, prefrontal cortex, cingulate cortex, and subcortical regions. This gradient underscores the biochemical integration of the brain’s network architecture. Furthermore, we demonstrate that structural hubs align with metabolically active nodes, reinforcing the relationship between network centrality and biochemical demand.
Importantly, we show that this metabolic network organization is not merely a byproduct of structural connectivity but is rooted in neurodevelopmental mechanisms, including genetic co-expression and cytoarchitectonic patterns. These results suggest that the topology of the metabolic connectome reflects underlying biological frameworks and developmental trajectories.
This work bridges the gap between metabolism and connectomics, offering a robust and replicable framework for constructing metabolic connectomes. By introducing the concept of the metabolic fiber and providing evidence for the biochemical basis of network organization, we lay the foundation for integrating MRSI into studies of brain health, disease, and neurodevelopment.