Diffusion MR Microscopy of the Human Hippocampus at 50 μm Resolution at 14.1 T for 3D Virtual Histology

Curation statements for this article:
  • Curated by eLife

    eLife logo

    eLife Assessment

    This paper presents important data from a very high-resolution MRI scan of the human hippocampus. The data, which is openly shared, convincingly shows that diffusion MRI at high resolution can reveal a lot of the known mesoscale anatomy. This study will be of interest to scientists who use neuroimaging to measure brain structure and function.

This article has been Reviewed by the following groups

Read the full article

Abstract

Bridging the gap between macroscopic neuroimaging and microscopic histology remains a challenge in neuroscience. Here, we present a technical feasibility study of diffusion MR microscopy (dMRM) in the ex vivo human hippocampus using a 14.1 T ultrahigh-field MRI system. Mid-hippocampal blocks from control and Alzheimer’s disease (AD) patients were acquired at 50 × 50 × 100 µm³, representing, to our knowledge, the highest resolution reported for human brain diffusion MRI. At a spatial scale approaching cellular dimensions, dMRM maps and tractography showed close correspondence with immunohistological markers of axons and dendrites. The acquired data delineated six hippocampal laminae, distinguished the inner and outer molecular layers, and preserved short, lamina-specific trajectories that became attenuated or blurred at 100 µm resolution. Tractography further depicted the major components of the hippocampal tri-synaptic circuit—the perforant path, mossy fibers, and Schaffer collaterals in 3D. In the AD specimen, dMRM revealed reduced laminar contrast, local discontinuities, and altered radial and tangential fibers that were paralleled by pathological staining. With the 3D nondestructive characterization of human brain microstructure and circuitry, ex vivo dMRM showed its potential toward the goal of 3D virtual histology.

Article activity feed

  1. eLife Assessment

    This paper presents important data from a very high-resolution MRI scan of the human hippocampus. The data, which is openly shared, convincingly shows that diffusion MRI at high resolution can reveal a lot of the known mesoscale anatomy. This study will be of interest to scientists who use neuroimaging to measure brain structure and function.

  2. Reviewer #1 (Public review):

    The authors present results from an ultra-high-resolution ex vivo MRI scanning of hippocampi. Scanning was extensive, and the figures are impressive. The description and scope of this resource appear a bit limited, however, as the data do not seem to be made openly available.

    (1) Title:

    The title is a bit ambiguous as "virtual histology" has also been used in an imaging-transcriptomics context (eg https://jamanetwork.com/journals/jamapsychiatry/fullarticle/2769908). I would suggest rewording or dropping everything after 14.1T.

    (2) Abstract:

    I would reword the statement "representing, to our knowledge, the highest resolution reported for human brain diffusion MRI" to "representing one of the highest resolutions reported
    for human brain diffusion MRI"

    (3) Introduction:

    When talking about selective vulnerability across disorders, the authors are encouraged to specifically mention AD, schizophrenia, and temporal lobe epilepsy, as these are conditions with a close association to hippocampal structural damage.

    (4) Introduction:

    It will, for balance, nevertheless be important in the introduction to outline that high-field in vivo MRI at 3-7T can achieve resolutions at least in the order of a few hundred microns. Such resolution, combined with automated image processing, can resolve subregional detail of the hippocampus. See for example https://pubmed.ncbi.nlm.nih.gov/31245529/, https://pubmed.ncbi.nlm.nih.gov/40654693/, https://www.nature.com/articles/s41592-025-02783-3 etc. As such, the statement that "clinical MRI fails to resolve the mesoscopic and microscopic organization essential for characterizing complex structures" is only partially correct, and I would suggest toning this down.

    (5) Methods:

    The methods appear relatively brief. The authors are encouraged to provide further details on the specific segmentation protocol that was used. Did multiple raters perform substructural segmentation? How high was inter- and intra-rater reliability?

    (5) Methods:

    How far is a diffusion MRI acquisition with 20 directions viable for tractography? I understand that multiple averages were used, but 20 directions appear limited to reliably sample 3D space and to, e.g., resolve complex, crossing, and kissing fibers.

    (6) Results:

    The figures are very nice, and the authors are also commended for doing such an ambitious imaging acquisition. Currently, the results appear a bit qualitative and descriptive, however. Ideally, and for greater impact, a resource such as the one presented should be made openly available for the research community, as this allows for independent verification and greater uptake of the study results. Other resources, e.g., AHEAD, BigBrain, etc. were shared with the international research community, so it would be great to see the same for this dataset.

  3. Reviewer #2 (Public review):

    This work uses extremely high-resolution diffusion MRI to reveal hippocampal layers and fiber organization in ex vivo tissue, showing compelling (and beautiful) evidence that this dMRI microscopy can truly act as "3D virtual histology" that gives information and signatures conveyed through a number of traditional stains.

    Specifically, they scan hippocampal tissue blocks (one from a control donor and one from an AD donor) using a 14.1T scanner, and compare diffusion maps and 3D tractography against axons, dendrites, amyloid, and tau. Acquisition is an impressive 50x50x100 µm resolution over 78 hours of scanning. They show (i) clearly distinguished hippocampal layers, (ii) this clear separation is lost at> 100 µm and beyond, (iii) tractography depicts several pathways consistent with the perforant path, mossy fibers, and Schaffer collaterals, (iv) the AD specimen showed weaker layer contrast, structural discontinuity, and disorganized trajectories (which agree with histological observations).

    Overall, I very much enjoyed reading this work; it read like a fully polished paper. I recommend only what I consider very minor revisions. And I would consider none a "requirement" for future revisions, as they are mostly stylistic preferences, small clarifications, and organizational choices.

    (1) The first results section "Diffusion MR microscopy protocol and multi-modal dataset of the human hippocampus" describes "methods". It contains/describes the acquisition, resolution, staining, and layer segmentation/naming (whereas the next section is the first results showing that diffusion microscopy allowed clear separation of layers).

    (2) There are some terms or scales that are introduced but may be ambiguous or not clear to readers. I'm assuming "diffusion MRI microscopy" is diffusion MRI on the scale of microns (or 10s of microns). "Meseoscopoic and microscopic" scales might be clarified.

    (3) "Structure tensor analysis" is introduced (and in the methods) without giving the reasons "why" this is used. The authors are using it to validate and compare orientation. The methods are described, but it isn't immediately intuitive what you are using ST analysis for (unless one is already familiar with this image processing technique).

    (4) This does not need to be addressed in the paper, but a recommendation: with 1mM Gd, you can likely get by with a much lower TR with 3D EPI, allowing much higher SNR-time efficiency, and it may be beneficial to avoid averaging and pursue more unique directions.

    (5) The image showing TDI might benefit from scaling each result independently - the 400 µm, by definition, has fewer streamlines per voxel, and is darker (and less visually appealing).

    (6) The Tri-synaptic pathway results are fascinating, a beautiful visualization of these small connections using tractography, and likely the first time I've seen this in a convincing manner using diffusion MRI. I'm unsure if the novelty of these findings is fully conveyed - adding a "virtual dissection" aspect to the "virtual histology".

    (7) Regarding Figure 5, "Disorganized radial dendritic orientations", are these disorganized, or rather geometrically different (i.e., less orthogonal), but still coherent? I wonder if, very simply, organizational coherence could additionally be quantified. Although, again, this is more a curiosity and specificity of the word "disorganized", and I do not consider further analysis a necessity.

    (8) Is the dataset made available (I may have missed this in the text)?