Algorithmic Geometry Shapes the Biological Meaning of Thalamocortical Structural Covariance

In 755 participants aged 8–22 from the Philadelphia Neurodevelopmental Cohort, PCA and OPNMF produced the strongest links between thalamocortical SC component weights and age.

• Why it matters: Structural covariance is widely used to infer brain organization, but the choice of algorithm may impose constraints that bias what neurobiology SC appears to reflect.
• What they did: Voxelwise thalamocortical SC networks from DBM were computed using PCA, ICA, NMF, OPNMF, and k-means, then compared with age-related individual differences, FA-based microstructural connectivity, and Allen Human Brain Atlas gene expression.
• The result: Algorithm choice shaped biological correspondence: PCA/OPNMF tracked age most, NMF/OPNMF/k-means aligned more with tractography projections, and ICA/PCA matched transcriptomic profiles while NMF/OPNMF/k-means more often highlighted thalamic components with high core-matrix gene expression.

AI summary of the abstract · may contain errors · not medical advice · read the paper.