The fixed samples were transferred to the Basque Center on Cognition, Brain and Language (BCBL, Donostia - San Sebastian, Spain) for MRI scanning. Ex vivo MR images of the whole brains were acquired on a 3 T Magnetom TIM Trio scanner with a 12 channel receiver coil. Despite its reduced efficiency compared with the 32 channel counterpart, the 12 channel coil enables acquisition at higher resolution without running out of RAM in the image reconstruction. The brains were scanned in vacuum bags filled with Fluorinert FC-3283 (3M, Maplewood, MN, U.S.A.), in order to minimize the negative impact of air bubbles and susceptibility artifacts. The images were acquired with a 3D multi-slab balanced steady-state free precession sequence (McNab et al., 2009 (link)) with TE/TR = 5.3/10.6 ms and flip angle 35 . Four axial slabs with 112 slices each were used to cover the whole volume of the brains, and 57% slice oversampling was used in order to minimize slab aliasing. The resolution of the scans was 0.25 mm isotropic, with matrix size 720 × 720 × 448 voxels (axial). MR images were acquired with four different RF phase increments (0, 90, 180, 270°) and averaged to reduce banding artifacts. The time of acquisition per phase was 90 min. Ten repetitions of this protocol were acquired for increased signal-to-noise ratio (SNR). The total length of the protocol was thus 60 h.
Combined with the 12 channel receiver coil, the multi-slab acquisition described above enabled us to bypass the memory limitations of our clinical scanner when reconstructing the images, while preserving the SNR efficiency of 3D acquisitions. However, this type of acquisition also introduces slab boundary artifacts at the interfaces between the slabs. After computing a brain mask with simple Otsu thresholding (Otsu, 1975 ), such artifacts were corrected simultaneously with the bias field using a Bayesian method (Iglesias et al., 2016 ). Sample slices of the MRI scans are shown in Fig. 1.

(a) Sample sagittal slice of ex vivo MRI scan of case NHL8_14. (b) Corrected for bias field and slab boundary artifacts. (c) Close-up of left thalami in coronal view, uncorrected. (d) Corrected version of (c).

Fig. 1
Free full text: Click here