Zhang · Human brain mapping 2022 · computational neuroimaging method validation study · n=?

Cerebral cortex layer segmentation using diffusion magnetic resonance imaging in vivo with applications to laminar connections and working memory analysis.

Level 5 - mechanism / opinion, no new human data

Methodological neuroimaging study validating a computational segmentation technique against reference histology

PubMed 35778791 · doi:10.1002/hbm.25998 · record verified 2026-08-26

What was done

The authors applied a K-means clustering algorithm to an open 7 Tesla (7T) diffusion magnetic resonance imaging (dMRI) dataset to automatically segment the left hemisphere cerebral cortex into superficial and deep layers in vivo. They validated the segmented layer thicknesses against histological reference data from the BigBrain dataset (which segments the neocortex into six layers based on the von Economo atlas). They subsequently constructed laminar connections for two pairs of unidirectionally connected brain regions and performed a laminar analysis of working memory.

What was found

The abstract reports a significant correlation between the dMRI-derived superficial layer thickness and histological layers 1–3, as well as between the dMRI-derived deep layer thickness and histological layers 4–6. Laminar connectivity between two pairs of unidirectionally connected regions matched prior literature. No exact numerical values (such as correlation coefficients, p-values, or segmentation error metrics) are reported in the abstract.

Why it matters

This study provides a noninvasive method for in vivo laminar segmentation of the human cortex using high-field dMRI, potentially facilitating future research on layer-specific connectivity and cognitive mechanisms.

Limits

The abstract does not state the participant sample size (n) or demographic details of the open dMRI dataset. Segmentation was restricted to the left hemisphere and resolved the cortex into only two broad compartments (superficial and deep) rather than all six histological layers. No quantitative statistics or validation metrics are provided in the abstract.