Visualizing myeloarchitecture with magnetic resonance imaging in primates

Abstract

The pattern of myelination over the cerebral cortex, termed myeloarchitecture, is an established and often-used feature to visualize cortical organization with histology in a variety of primate species. In this paper, we use in vivo magnetic resonance imaging (MRI) and advanced image processing using surface rendering to visualize and characterize myeloarchitecture in a small nonhuman primate, the common marmoset (Callithrix jacchus). Through images made in four female adult marmosets, we produce a representative 3D map of marmoset myeloarchitecture and flatten and annotate this map to show the location and extent of a variety of major areas of the cortex, including the primary visual, auditory, and somatosensory areas. By treating our MRI data as a surface, we can measure the surface area of cortical areas, and we present these measurements here to summarize cortical organization in the marmoset.

Figure 1

Representative cortical myelination in the marmoset. A 40 μm-thick coronal histological section stained for myelin using a modified Gallyas silver staining method. At the whole mount level of magnification showing half of the brain, distinct dense areas of staining are seen which correspond to specific cortical areas (for example, the primary auditory cortex (A1)). At 2.5 times magnification, dark stained fibres can been seen running vertically from the white matter (WM) through Layers VI and V. At 20 times magnification in Layer IV, the fibres are arborized and vertical and horizontal stained branches can be seen. There is little myelin staining through Layers III -I. The same pattern is seen in other myelinated areas of the marmoset cortex, save for the primary visual cortex (V1) where the density of myelination is highest in Layer IV (the Stripe of Gennari). The asterix denotes the same blood vessel at each magnification.

Figure 2

Image Processing. A 40 μm-thick representative coronal histological section stained for myelin (Panel 1) and a matching 165 μm thick corresponding T₁-weighted MRI section from a 3D image (Panel 2). The density of myelination is generally highest in a middle depth of the cortex (around Layer VI); thus, we define our surface at this depth (Panel 3). The 3D surface over the entire cortex at a middle depth is shown using surface rendering (Panel 4). The final step in our processing is to project the local MRI image intensity onto this surface to make a map.

Figure 3

3D map of myeloarchitecture in a representative 3-year old female common marmoset. The figure shows a view of a 3D map of the cortex in a marmoset centered on the dorsal parietal cortex (Panel 1). Here, the MRI intensity data is displayed using a hot colourmapto highlight contrast and areas of enhancement represent cortical areas with high myelin contents. The map is placed at a middle depth in the cortex, and the surface corresponding to the outside of the cortex is shown in light transparent blue. By rotating the map to a view centered on the occipital cortex (Panel 2), we can better see the primary visual cortex (V1) and by making the dorsal surface of our map transparent (shown in red), we can see the extension of V1 into the calcarine fissure. (C = caudal, R = rostral, V = ventral, D = dorsal)

Figure 4

Flattened map of myeloarchitecture in a representative 3-year old female common marmoset. The dorsal cortical surface from the map in Figure 3 is flattened with the major enhancing areas labeled. Note that the flattening produces spatial distortions in the surface (see Supplemental Figure 1 for details); however, the flattening allows for easier visualization of cortical features and their spatial relationships. “Low SNR” denotes an MRI artifact where poor radiofrequency coil coverage led to a non-specific image enhancement. This was because the RF coil did not provide coverage rostrally completely to the frontal pole This distinct area of low SNR was readily apparent during our corrections for B₁ inhomogeneity in the images.(C = caudal, R = rostral, L = lateral, M = medial). (Major myelinated cortical areas are labeled in white: V1 = primary visual area, MT = middle temporal area, A1 = primary auditory area, R = rostral auditory area, S1 = primary somatosensory cortex, M = motor cortex including primary and premotor areas and the frontal eye fields). (Cortical features are labeled in gray: DM = dorsomedial area, PPv = ventral posterior parietal cortex, FST = fundus of the superior temporal area, S2 = secondary somatosensory cortex, PV= parietal ventral area, 12 = area 12).

Figure 5

Comparative 3D maps of myeloarchitecture from representative 3-year old and 8-year old female common marmosets. Both animals are adults and the maps show a similar pattern of cortical organization.

Figure 6. Definition of cortical areas for surface area measurements

Figure 7

Areal distortion introduced by flattening. The figure shows the areal distortion produced when the right hemisphere was flattened in Figure 4. The original surface can be represented as a mesh of triangles. The areal distortion is a localized measure of the expansion or contraction of these triangles produced by the flattening operation. It is defined as ADi = log2(ASi/AFi) where i indexes a given triangle, ASi is its area in the original surface and AFi is its area in the flattened surface. In the figure, areal distortions tending towards black represent shrinkage, while those tending towards white represent expansions. One can see that the distortion is not uniform over the cortex, and it is greatest in regions with a high degree of curvature in the original image, like the frontal pole.

Figure 8

Views of the surface rendered images from each of the four marmosets in the study.