High resolution diffusion imaging in the unfixed post-mortem infant brain at 7T

Abstract

Diffusion MRI of the infant brain allows investigation of the organizational structure of maturing fibers during brain development. Post-mortem imaging has the potential to achieve high resolution by using long scan times, enabling precise assessment of small structures. Technical development for post-mortem diffusion MRI has primarily focused on scanning of fixed tissue, which is robust to effects like temperature drift that can cause unfixed tissue to degrade. The ability to scan unfixed tissue in the intact body would enable post-mortem studies without organ donation, but poses new technical challenges. This paper describes our approach to scan setup, protocol optimization, and tissue protection in the context of the Developing Human Connectome Project (dHCP) of neonates. A major consideration was the need to preserve the integrity of unfixed tissue during scanning in light of energy deposition at ultra-high magnetic field strength. We present results from one of the first two subjects recruited to the study, who died on postnatal day 46 at 29+6 weeks postmenstrual age, demonstrating high-quality diffusion MRI data. We find altered diffusion properties consistent with post-mortem changes reported previously. Preliminary voxel-wise and tractography analyses are presented with comparison to age-matched in vivo dHCP data. These results show that high-quality, high-resolution post-mortem data of unfixed tissue can be acquired to explore the developing human brain.

Fig. 1.

Simulation of accelerations for different protocols. (a) Predicted SAR for different acceleration protocols with minimum allowable TRs. In order to run a protocol consecutively over many hours, the sequence SAR needs to be lower than the conservative SAR limit (black dotted line). Burst mode allows a sequence to run at full SAR limit for a maximum of 6 minutes (red dot line). Protocols with b = 9000 s/mm² (blue solid line) have a lower SAR than the b = 4000 s/mm² protocols (red solid line) due to the long TRs required for large b values. (b) Minimum allowable TRs for all protocols under investigation. To be SAR compatible, some protocols with high MB factors need to use a higher TR (dash line). (c) The relation between the achievable number of diffusion volumes with TRs. In order to acquire at least 301 image volumes in the 7-hour scan, the TR should not exceed 12 s (dash line), which excludes five protocols including all MB = 1 protocols and MB2R1 (i.e., MB factor 2 with in-plane phase encoding acceleration R = 1) protocol.

Fig. 2.

Simulation of blurring and distortion. (a) Blurring simulation: plot of voxel point spread functions along phase encoding direction for different phase encoding acceleration factors with an estimated T2* of 83 ms. (b) Distortion simulation: histogram of voxel shift distances (mm) simulated based on a representative B0 field map that was acquired at 3 T from a healthy infant and linearly scaled to 7 T.

Fig. 3.

Evaluation of reconstruction performance for different acceleration protocols. (a) Reconstructed phantom images for different combinations of MB acceleration factors and phase encoding acceleration factors R. Normalized root mean square error is shown on the bottom right corner of each image. (b) Histogram of g-factors over all voxels for MB2R2, MB3R2, MB4R2, and MB2R3. µ = median g-factor.

Fig. 4.

(a) Temperature of brain tissue (red) and ambient (blue) during a 6-hour diffusion scan with a porcine brain without active cooling. Active cooling was applied during data acquisitions for (b) and (c). (b) Temperature measured from the inside (red) and surface (blue) of a tissue-mimicking phantom. (c) Temperature measured during a post-mortem dHCP scan, with two probes placed on the forehead (blue) and underarm (red) of the infants. The target temperature 8 °C set for the cooler in the post-mortem dHCP scan is shown as a dotted horizontal line. (d) Median ADC of brain tissues calculated from the eight sub-groups of b = 9000 s/mm² post-mortem dHCP data.

Fig. 5.

Post-mortem dHCP images from one of the first two subjects. Three b shells (3000, 6000, and 9000 s/mm²) were acquired at 0.8 mm isotropic resolution. Here the b = 0 images and shell-averaged diffusion weighted images are shown. Note this infant had bilateral grade 3 intraventricular hemorrhages (IVH) and post-hemorrhagic ventricular dilatation, as can be observed from the images.

Fig. 6.

Assessment of diffusion kurtosis model fit. There is a high degree of qualitative correspondence in spatial patterns between the post-mortem data and the exemplar age-matched in vivo subject. In general, greater microstructural detail can be resolved in the post-mortem data relative to the in vivo data due to the higher spatial resolution. Note the acquisition time for the post-mortem data is much longer than the in vivo data (7 hours vs 19 minutes). Principal diffusion direction from DKI model fit of post-mortem (a) and in vivo (c) dHCP data. FA (fractional anisotropy), MO (mode of the anisotropy), MD (Mean diffusivity, expressed in µm²/ms), and MK (mean kurtosis) parameters from the DKI model fit of post-mortem (b) and in vivo (d) dHCP data. Quantitatively, overall spatial patterns of DKI parameters are consistent between subjects. However, in the post-mortem data, the MD values are noticeably lower, while the MK values are noticeably higher, than the in vivo data. Note the change of color bar scale between the post-mortem and in vivo maps. The post-mortem data has a resolution of 0.8 mm isotropic, and the in vivo data have a resolution of 1.5 mm isotropic (but were interpolated to 1.17 x 1.17 x 1.5 mm³ in post-processing).

Fig. 7.

Assessment of NODDI model fit results. The parameter maps are “Intra-neurite volume fraction”, “isotropic volume fraction”, and “orientation dispersion” index parameters from NODDI model fit of post-mortem and the exemplar age-matched in vivo subject. Overall, the spatial patterns of NODDI parameters are quantitatively consistent between subjects. However, in the post-mortem data, the “intra-neurite volume fraction” values are noticeably higher than the in vivo data. Note the change of color bar scale between the post-mortem and in vivo maps. The post-mortem data have a resolution of 0.8 mm isotropic, and the in vivo data have a resolution of 1.5 mm isotropic (but were interpolated to 1.17 x 1.17 x 1.5 mm³ in post-processing).

Fig. 8.

Axial (a, d), sagittal (b, e) and coronal (c, f) sections of white matter ODFs overlaid on CSF components (background) for post-mortem dHCP data and data from the age-matched in vivo dHCP subject. Abbreviations: cc = corpus callosum; atr = anterior thalamic radiation; cst = corticospinal tract; cgc = cingulate gyrus part of the cingulum; tpf = transverse pontine fibers.

Fig. 9.

Maximum intensity projections of 11 tracts from the in vivo and post-mortem data. All tracts are visualized using the same thresholds (0.005, 0.05). Abbreviations: ar = acoustic radiation; atr = anterior thalamic radiation; cgc = cingulate gyrus part of the cingulum; cgh = parahippocampal part of the cingulum; cst = corticospinal tract; for = fornix; ilf = inferior longitudinal fasciculus; ptr = posterior thalamic radiation; slf = superior longitudinal fasciculus; str = superior thalamic radiation; unc = uncinate fasciculus.