Automated Parcellation of Human Cerebral Cortex Using Gross Morphological Characteristics

IntroductionBrodmann [۱] identified ۵۵ areas on the human cortex using cytoarchitecture patterns and morphological differences between cortical regions. However, no quantitative method was used. Moreover, several studies have used functional and structural imaging techniques that suggest more cortical areas. In this study, human factor was removed and mathematical measurements were applied on the BigBrain dataset to parcellate human cerebral cortex.MethodFirstly, the pial layer and the gray matter / white matter boundary of the ۸bit ۲۰۰μm BigBrain were identified using Freesurfer software to measure five features: thickness, surface area, volume, curvature, and coordination for every vertex. Secondly, all the vertexes were distributed to ۳۶ blocks based on the Desikan-Killiany atlas. Finally, a density-based scan (DB-scan) clustering algorithm [۲] was applied on features and the input parameters were obtained using mathematical measurements applied on the data.ResultsAfter processing the data, ۱۵۶ and ۱۵۹ areas for the left and the right hemisphere were identified, respectively (Figure ۱) with a RGB color for each one. The coordination feature was multiplied to ۱۰ to enhance the integration of the areas, and thickness was multiplied to ۴۰ for more significant results. To mention some important results, as illustrated in figure ۲, supplementary motor area(SMA) was identified on the first sections on the left hemisphere (۲B), but in deeper sections in the right hemisphere (۲C, ۲E). Pre-SMA was identified in deep sections on both hemispheres (۲B, ۲E). Finally, the primary motor cortex and premotor cortex were parcellated into smaller areas (۲A, ۲C, ۲D).