7 Tesla MRI of the ex vivo human brain at 100 micron resolution

Abstract

We present an ultra-high resolution MRI dataset of an ex vivo human brain specimen. The brain specimen was donated by a 58-year-old woman who had no history of neurological disease and died of non-neurological causes. After fixation in 10% formalin, the specimen was imaged on a 7 Tesla MRI scanner at 100 µm isotropic resolution using a custom-built 31-channel receive array coil. Single-echo multi-flip Fast Low-Angle SHot (FLASH) data were acquired over 100 hours of scan time (25 hours per flip angle), allowing derivation of synthesized FLASH volumes. This dataset provides an unprecedented view of the three-dimensional neuroanatomy of the human brain. To optimize the utility of this resource, we warped the dataset into standard stereotactic space. We now distribute the dataset in both native space and stereotactic space to the academic community via multiple platforms. We envision that this dataset will have a broad range of investigational, educational, and clinical applications that will advance understanding of human brain anatomy in health and disease.

Fig. 1

Human brain specimen. The human brain specimen that underwent ex vivo MRI is shown from inferior (a), superior (b), right lateral (c) and left lateral (d) perspectives. Gross pathological examination of the brain was normal.

Fig. 2

Receive array coil and transmit volume coil for ex vivo imaging of the whole human brain. (a) The 31-channel receive array has 15 elements on the top half (with a diameter of 5.5 cm) and 16 on the bottom half (with a diameter of 8.5 cm), each made of 16 AWG wire loops with four or five evenly spaced capacitors. All elements are tuned to 297.2 MHz. (c) The coil former has slightly larger dimensions than the brain holder, which slides inside a volume coil (b). (d) A custom air-tight brain holder was designed to conform to the shape of a whole human brain. The brain holder is an oblate spheroid container (16 × 19 cm) with degassing ports that are used to apply a vacuum suction, thereby reducing air bubbles in the specimen and surrounding fomblin solution.

Fig. 3

Comparison of FA25° acquisition and synthesized FLASH25 volume. Representative images from the FA25° acquisition (left column) and the synthesized FLASH25 volume (right column) are displayed in the sagittal (top row), coronal (middle row) and axial (bottom row) planes. These images provide a qualitative comparison of the respective signal-to-noise properties of the FA25° acquisition (~25 hours) and the synthesized FLASH25 volume (~100 hours). All images are shown in radiologic convention.

Fig. 4

Delineation of brainstem neuroanatomy. Representative axial sections from the synthesized FLASH25 volume are shown at the level of the rostral pons and caudal midbrain (a-c, see inset in panel c). Zoomed views of the brainstem, medial temporal lobe, and anterior cerebellum (within the white rectangles in a–c) are shown in the bottom row (d–f). The anatomic detail that can be visualized in this ex vivo 100 μm resolution MRI dataset is beyond that which can be seen in typical in vivo MRI datasets. All images are shown in radiologic convention. Neuroanatomic abbreviations: Amg = amygdala; Cb = cerebellum; CP = cerebral peduncle; MB = mammillary body; P = pons; SCP = superior cerebellar peduncle; VTA = ventral tegmental area; xSCP = decussation of the superior cerebellar peduncle; Th = thalamus.

Fig. 5

Delineation of basal ganglia and basal forebrain neuroanatomy. A representative coronal section from the synthesized FLASH25 volume is shown in the plane of the anterior commissure (aComm; see inset in a). A zoomed view of the basal ganglia and basal forebrain (within the white rectangle in a) is shown in (b). The anatomic detail that can be visualized in this ex vivo 100 μm resolution MRI dataset is beyond that which can be seen in standard in vivo MRI datasets. Neuroanatomic abbreviations: C = caudate; CB = cingulum bundle; CC = corpus callosum; Cl = claustrum; Fx = fornix; GPe = globus pallidus externa; IC = internal capsule; NBM = nucleus basalis of Meynert; Ox = optic chiasm; Put = putamen; Sb = striatal bridges.

Fig. 6

Signal-to-noise ratio (SNR) analysis of coil performance. Representative SNR maps are shown in the sagittal (top row), coronal (middle row) and axial (bottom row) planes for a test brain sample immersed in periodate-lysine-paraformaldehyde. The maps demonstrate an SNR gain of 1.6-fold for the 31-channel 7 Tesla (7 T) ex vivo coil (left column) compared to the 31-channel 7 T standard coil (middle column), and a gain of 3.3-fold compared to the 64-channel 3 T head coil (right column). The noise coupling between channels was 11% for the 31-channel ex vivo coil array, a 2-fold improvement relative to our previous array.

Fig. 7

Normalization of the ex vivo MRI dataset into standard stereotactic space and integration into the Lead-DBS software platform. (a) Exemplary use-case of the normalized FLASH25 volume in deep brain stimulation (DBS). DBS electrodes are visualized for a hypothetical patient using Lead-DBS software (https://www.lead-dbs.org). An axial image from the normalized scan, at the level of the rostral midbrain, is shown as a backdrop, with 3D-structures defined by the DISTAL atlas (right subthalamic and left red nucleus hidden for optimal visualization of the underlying anatomy). Panels (b) and (c) show zoomed views of key DBS target regions: the left globus pallidus interna (GPi in b) and subthalamic nucleus (STN in c).