Ultra-high resolution diffusion tensor imaging of the microscopic pathways of the medial temporal lobe

Abstract

Diseases involving the medial temporal lobes (MTL) such as Alzheimer's disease and mesial temporal sclerosis pose an ongoing diagnostic challenge because of the difficulty in identifying conclusive imaging features, particularly in pre-clinical states. Abnormal neuronal connectivity may be present in the circuitry of the MTL, but current techniques cannot reliably detect those abnormalities. Diffusion tensor imaging (DTI) has shown promise in defining putative abnormalities in connectivity, but DTI studies of the MTL performed to date have shown neither dramatic nor consistent differences across patient populations. Conventional DTI methodology provides an inadequate depiction of the complex microanatomy present in the medial temporal lobe because of a typically employed low isotropic resolution of 2.0-2.5 mm, a low signal-to-noise ratio (SNR), and echo-planar imaging (EPI) geometric distortions that are exacerbated by the inhomogeneous magnetic environment at the skull base. In this study, we pushed the resolving power of DTI to near-mm isotropic voxel size to achieve a detailed depiction of mesial temporal microstructure at 3 T. High image fidelity and SNR at this resolution are achieved through several mechanisms: (1) acquiring multiple repetitions of the minimum field of view required for hippocampal coverage to boost SNR; (2) utilizing a single-refocused diffusion preparation to enhance SNR further; (3) performing a phase correction to reduce Rician noise; (4) minimizing distortion and maintaining left-right distortion symmetry with axial-plane parallel imaging; and (5) retaining anatomical and quantitative accuracy through the use of motion correction coupled with a higher-order eddy-current correction scheme. We combined this high-resolution methodology with a detailed segmentation of the MTL to identify tracks in all subjects that may represent the major pathways of the MTL, including the perforant pathway. Tractography performed on a subset of the data identified similar tracks, although they were lesser in number. This detailed analysis of MTL substructure may have applications to clinical populations.

Figure 1

Diffusion-weighted imaging pre-processing pipeline. See 2.1.2.

Figure 2

Medial temporal lobe neuroanatomy. At the top left is a color-coded diagram of medial temporal circuitry. Subregions are labeled on the overlaid coronal MR at the top right. Crosshairs denote tracks that run through plane. Pathways of interest: (1) parahippocampal cingulum bundle, which runs almost exclusively through-plane, (2) perforant (internal) and alvear (external) pathways, (3) CA3 and subicular projections to the fornix, (4) Schaffer collaterals from CA 3 to CA 1, (5) CA 1 to subiculum, (6) subiculum to entorhinal cortex, (7) entorhinal to perirhinal/parahippocampal cortex.

Figure 3

Segmentation Procedure: Coronal slabs 1–6 progress anteriorly to posteriorly on this left hemisphere. Slabs 1–4 constitute the hippocampal head, slab 5 represents the hippocampal body, and slab 6 represents the posterior-most hippocampal body. All images are isoDWI except for the far right column, which is a co-planar fractional anisotropy map.

Figure 4

Example high-resolution images of the right hippocampus from one subject. A) coronal isoDWI. B) coronal fractional anisotropy map. C) diffusion tensor ellipsoids. D) direction of principle eigenvector.

Figure 5

Tractography Parameter Optimization. In one subject, tracts are depicted with anisotropy thresholds of 0.02, 0.05, and 0.1 as well as minimum tract lengths of 10 and 20 mm, all utilizing a curvature threshold of 40 degrees. White arrows correspond to false positives transgressing CSF. Blue arrows correspond to a track seen only on the shorter minimum tract length threshold.

Figure 6-1

Parahippocampal cingulum bidirectional tractography in all 6 subjects. The color scale for the fiber pathways and subregions corresponds to that used in Figures 2 and 3, respectively.

Figure 6-2

Perforant pathway between the entorhinal cortex and CA3DG, CA 1, and SRLMHS. Perforant pathway fibers should project through the subiculum (which is also shown but is neither a seed nor a target for this pathway), while the alvear path fibers should extend around the hippocampus.

Figure 6-3

CA3 and subicular projections to the fornix.

Figure 6-4

Schaffer collaterals from CA 3 to CA 1/SRLMHS

Figure 6-5

CA 1 to subiculum