A new study from the Mark and Mary Stevens Neuroimaging and Informatics Institute at the Keck School of Medicine of USC, published in Nature Communications, offers a way to answer a deceptively simple question that has long quietly troubled neuroimaging researchers: does the brain's blood supply actually track where its cells need energy most? The team, reporting on work led by co-first authors Fanhua Guo and Chenyang Zhao, introduces the cerebral blood flow–cell-body staining intensity similarity index — CCSI — a measure designed to quantify how closely perfusion patterns align with the cellular architecture of the cortex. Consider the implications for anyone who builds or validates MRI pipelines. The brain has almost no ability to store energy, so its cells depend on a constant and carefully regulated supply from the bloodstream, and CCSI gives researchers a way to study how well that energy supply is positioned to meet cellular demands in different parts of the cortex.
How the measurement was built
The method rests on arterial spin labeling, a noninvasive MRI technique that magnetically labels water in the blood and tracks it into brain tissue. Here the team pushed the technique onto a 7 Tesla scanner, measuring perfusion across the brain at a resolution of one cubic millimeter — fine enough to resolve cortical layers rather than collapsing them into a single average. The study included 30 healthy adults, with 14 returning for a second scan to test the consistency of the measurements across time.
Rather than treating the cortex as a uniform sheet, the researchers divided it into 360 regions and examined blood flow at multiple depths. They then compared those perfusion patterns against cell-body staining data from BigBrain, a detailed three-dimensional digital reconstruction of a human brain that maps how densely cells are packed throughout the cortex. The resulting CCSI score reflects how closely blood flow and cellular density follow the same trajectory across cortical depth; higher scores indicate that cell-dense layers tend to receive proportionally more blood flow.
This is where the approach begins to feel genuinely useful for software engineers and protocol developers. Conventional brain imaging often averages information across the full thickness of the cortex, but the cortex is not a uniform sheet. By imaging blood flow at very high resolution, the team can begin to see how perfusion changes from the outer surface of the cortex to its deeper layers — a distinction that may matter enormously when modeling regional vulnerability to ischemia or to the subtle degradation that accompanies aging.
What this opens up
Postdoctoral scholar Ravi R. Bhatt frames the broader shift with care: neuroscience is entering an era where we can combine advanced brain imaging with detailed maps of the brain's cells, gene expression, and metabolism. Each map provides a different view of the brain, and together they reveal patterns that would otherwise remain hidden. CCSI sits at exactly that intersection — a similarity index that lives between vascular imaging and cytoarchitectonic reference data, asking the cortex to speak in layers rather than in averages.
For the moment, the findings are anchored in 30 healthy adults, which means the disease-related applications are still prospective. The authors position CCSI as a potential window into early signs of disease and possible treatment response, though any clinical claims remain firmly in the future tense. What is concrete today is a reproducible, layer-resolved way to ask whether perfusion and cellular density stay coupled across the cortex — and a methodology that engineers working on ASL pipelines can begin to test against their own acquisition protocols.
If subsequent work extends CCSI into aging cohorts, vascular disease, or the earliest stages of neurodegeneration, this similarity index could become one of those quiet but consequential tools that shifts how we read a brain scan. Until then, it is a careful proof of concept, and a reminder that the most informative imaging measures are often the ones that ask the cortex to speak in layers, not in averages.