Laminar signal extraction over extended cortical areas by means of a spatial GLM

摘要: There is converging evidence that distinct neuronal processes leave distinguishable footprints in the laminar BOLD response. However, even though achievable spatial resolution functional MRI has much improved over years, it still challenging to separate signals arising from different cortical layers. In this work, we propose a new method extract signals. We use General Linear Model combination with equivolume principle of layers unmix instead interpolating through and integrating area: thus reducing partial volume effects. Not only do provide mathematical framework for extracting GLM, also illustrate best case scenarios existing methods can be seen as special cases within same framework. By means simulation, show approach sharper point spread function, providing better signal localisation. further assess contamination profiles high human ex vivo structural data, full account benefits potential caveats. eschew here any attempt validate GLM on basis fMRI data generally accepted ground-truth pattern activation does not currently exist. This flexible terms number their respective thickness, naturally integrates regularisation along cortex, while preserving specificity. Care must taken, however, procedure unmixing susceptible sources noise or inaccuracies segmentation.