Cellular-resolution X-ray microtomography of an entire mouse brain: supporting data, scripts and parameters

Abstract

Purpose: Histology is the gold standard for sub-cellular visualization of the mouse brain. It offers excellent in-plane resolution, but a comparably low out-of-plane resolution due to physical sectioning. X-ray microtomography does not require this trade-off. Tomographic imaging of the entire mouse brain with isotropic cellular resolution produces datasets of multiple terabytes in size. These data must be processed and made accessible to domain experts who may have only limited image processing knowledge.

Approach: Extended-field X-ray microtomography covering an entire mouse brain was performed. The 4,495 projections from 8 × 8 offset acquisitions were stitched to reconstruct a 15,000^3 voxel volume. The microtomography volume was non-rigidly registered to the Allen Mouse Brain Common Coordinate Framework v3 based on a combination of image intensity and landmark pairs.

Results: We present a 3.3 teravoxel dataset covering a full mouse brain with 0.65 μm voxel size. The data were block-wise transformed to a common coordinate system then stored in a public repository with a hierarchical format for navigation and overlay with anatomical annotations in online viewers such as Neuroglancer or siibra-explorer.

Conclusions: This study demonstrates X-ray imaging and data processing for a full mouse brain, augmenting current atlases by improving resolution in the third dimension by an order of magnitude. The data are publicly available and easily accessible for domain experts via browser-based viewers.

Data

File b32_atlas_registration.zip contains the input data, parameter files and output data of affine and non-rigid registration of the 32 x 32 x 32 voxels downsampled microtomography image to the Allen Mouse Brain Common Coordinate Framework v3.

File create_sharded_Neuroglacer_format.pdf will provide instructions how to create gZip-compressed, sharded, pre-computed Neuroglancer format via the open source software igneous. This is currently still missing.

Processing

Reconstruction of extended field microtomography images mit projection stitching in horizontal and vertical direction was achieved with our software pipeline mosaicreconstruction.

The 32 x  32 x 32 voxels downsampled microtomography image was registered with the open source software elastix (and transformix). The elastix.log file states the used command line options. Directory names require adapting for reproducing the results. 

As terabyte large images and the non-rigid transformation parmeters do not fit into memory, we developed a distributed hierarchical approach with sub-volume transformations for the high resolution level, see LargeVolumeTransformix. The high resolution tomography volume (moving volume) was warped to the atlas (fixed volume) by dividing the fixed space into 12 × 12 × 12 target subregions, such that the axes-aligned extended moving image subregion, the moving image transformed subregion, and the transformation parameters all fit in memory for an isotropic target voxel resolution of 0.65 μm.