Direct segmentation of cortical cytoarchitectonic domains using ultra-high-resolution whole-brain diffusion MRI

Area of Science:

• Neuroimaging
• Computational Neuroscience
• Primate Brain Anatomy

Background:

• Conventional cortical parcellation relies on template-warping methods, which can be imprecise for delineating fine-grained anatomical features.
• Understanding the three-dimensional cytoarchitectural organization of the cortex is crucial for mapping brain function and connectivity.
• High-resolution imaging techniques are needed to capture the detailed microstructural variations within the cerebral cortex.

Purpose of the Study:

• To evaluate the efficacy of clustering microstructural parameters from high-resolution mean apparent propagator (MAP) magnetic resonance imaging (MRI) for detecting cortical laminar patterns and areal borders.
• To compare the accuracy of MAP-MRI-based cytoarchitectonic segmentation with conventional template-warping methods and histological data.
• To explore the potential of MAP-MRI for constructing whole-brain mesoscopic cortical atlases.

Main Methods:

• Acquisition of high-resolution (200 μm) MAP-MRI data from a fixed macaque monkey brain.
• Application of a local anisotropic Gaussian filter, optimized voxel-wise, to enhance sensitivity to cortical layers.
• Direct clustering of all cortical voxels based solely on MAP-derived microstructural biomarkers, without spatial location information.

Main Results:

• MAP-based 3D cytoarchitectonic segmentation revealed laminar patterns consistent with histological findings.
• The transition regions identified by MAP-MRI more accurately matched histological borders than those from conventional atlas-based parcellation.
• Automatic cross-hemispheric matching of MAP-derived cytoarchitectonic domains was achieved by cross-tabulating atlas- and MAP-based segmentations.

Conclusions:

• High-resolution MAP-MRI biomarkers effectively delineate three-dimensional cortical cytoarchitectonic domains within individual subjects.
• This direct microstructural clustering approach provides a more accurate alternative to template-warping for cortical parcellation.
• The intrinsic microstructural contrasts in MAP-MRI facilitate the creation of comprehensive whole-brain mesoscopic cortical atlases.