A framework for cortical laminar composition analysis using low-resolution T1 MRI images

Abstract

The layer composition of the cerebral cortex represents a unique anatomical fingerprint of brain development, function, connectivity and pathology. Historically the cortical layers were investigated solely ex-vivo using histological means, but recent magnetic resonance imaging (MRI) studies suggest that T1 relaxation images can be utilized to separate the layers. Despite technological advancements in the field of high resolution MRI, accurate estimation of whole brain layer composition has remained limited due to partial volume effects, leaving some layers far beyond the image resolution. In this study we offer a simple and accurate method for layer composition analysis, resolving partial volume effects and cortical curvature heterogeneity. We use a low resolution echo planar imaging inversion recovery (EPI IR) MRI scan protocol that provides fast acquisition (~12 minutes) and enables extraction of multiple T1 relaxation time components per voxel, which are assigned to types of brain tissue and utilized to extract the subvoxel composition of each T1 layer. While previous investigation of the layers required the estimation of cortical normals or smoothing of layer widths (similar to VBM), here we developed a sphere-based approach to explore the inner mesoscale architecture of the cortex. Our novel algorithm conducts spatial analysis using volumetric sampling of a system of virtual spheres dispersed throughout the entire cortical space. The methodology offers a robust and powerful framework for quantification and visualization of the layers on the cortical surface, providing a basis for quantitative investigation of their role in cognition, physiology and pathology.

Modelling cortical laminar connectivity in the macaque brain

“…This sequence was used as an anatomical reference with high gray/white matter contrast and used for segmentation and inner and outer surface estimation. Inversion recovery was performed using a 3D FLASH sequence with: voxel size 0.67×0.67×0.67 mm3, image size 96×96×68 voxels, TR/TE=1300/4.672 ms and 44 inversion times spread between 25 ms up to 1,000 ms, each voxel fitted with up to 8 T1 values (similarly to Lifshits et al 2018). T1w and inversion recovery datasets were used for cortical laminar composition analysis (similarly to the framework presented in Shamir et al 2019). DWI (diffusion weighted imaging) was performed with 4 segments diffusion weighted EPI sequence with a voxel size 0.48×0.48×0.48 mm3, image size 128×160×116 voxels, Δ/δ=20/3.3 ms, b=5000 s/mm2, with 96 gradient directions and additional 4 with b=0. DWI dataset was used for global white matter connectivity (tractography), using constrained spherical deconvolution (CSD) in ExploreDTI (Leemans et al 2009).…”

“…Inversion recovery was performed using a 3D FLASH sequence with: voxel size 0.67×0.67×0.67 mm3, image size 96×96×68 voxels, TR/TE=1300/4.672 ms and 44 inversion times spread between 25 ms up to 1,000 ms, each voxel fitted with up to 8 T1 values (similarly to Lifshits et al 2018). T1w and inversion recovery datasets were used for cortical laminar composition analysis (similarly to the framework presented in Shamir et al 2019).…”

“…In order to provide accurate volumetric sampling of the complex geometry of the cortex, we use the unique framework for cortical laminar composition analysis (presented by Shamir et al 2019). The framework uses a sampling system made up of virtual cortical spheres, as a symmetric and rotationally invariant volumetric alternative to cortical normals.…”

“…In recent years, promising progress has also been made in the field of MRI imaging of the cortical laminar composition of the cortex (Barazany and Assaf 2012, Lifshits et al 2018, Shamir et al 2019). More specifically, the methodology (presented by Shamir et al 2019) offers a framework for MRI-based quantification and visualization of the cortical laminar composition.…”

“…In recent years, promising progress has also been made in the field of MRI imaging of the cortical laminar composition of the cortex (Barazany and Assaf 2012, Lifshits et al 2018, Shamir et al 2019). More specifically, the methodology (presented by Shamir et al 2019) offers a framework for MRI-based quantification and visualization of the cortical laminar composition. By integrating the laminar structure derived from this framework into connectomics, not only we will be able to give a more detailed and unbiased representation of brain connectivity, but we will also open a new avenue to future prospects in the field of connectomics (see figure 1).…”