Importing and modifying the FP labels to the Allen CCF

We originally obtained vector drawings of Nissl 2D section from Paxinos and Franklin’s the Mouse Brain in Stereotaxic Coordinates, 3rd edition⁶. We also used the 4th version to incorporate the latest updated labels. We used a vector drawing tool (Adobe Illustrator) for our labeling work. We downloaded the Allen CCF and associated labels from the Allen Institute for Brain Sciences API (http://help.brain-map.org/display/mousebrain/API), and generated coronal slices (10 μm isotropic) using Image-Stacks-Reslices in FIJI (NIH)⁶². This produced 1320 Z coronal slices. Then, we selected one coronal slice in every 10 slices from Z95 to Z1315 using Image-Stacks-Tools-Make Substack in FIJI, generating 123 coronal images with 100 μm z spacing. We chose 100 μm z spacing to facilitate comparisons between the ARA and the CCFv3 because the ARA was created in coronal sections evenly spaced at 100 μm intervals³⁸. We identified matching z planes between the FP atlas and the CCF using distinct anatomical landmarks (e.g., fiber track, and ventricles). Anterior-posterior (A/P) Bregma coordinates of z sections were primarily based on ARA⁷ while cross-referencing to the FP atlas⁶. To aid our label alignment in 3D, we downloaded MRI labels from different brain regions from a publically available database (https://imaging.org.au/AMBMC/AMBMC). We combined labels from different brain regions to reconstruct the MRI labels using FIJI (NIH)⁶². Then, we registered the MRI atlas with the FP based labels to the CCF using Elastix. The MRI labels were particularly useful to align boundaries in cortical areas in 3D. We loaded cell type specific labeling from different transgenic mice and MRI labels as separate layers on the Illustrator, and used the information to further adjust anatomical delineations. To accommodate the FP labels (mostly 120 μm z spacing) in 100 μm z spacing, we used the 5th section of every 6 FP labels twice in the initial alignment and used the MRI atlas and marker brains to further modify the labels across the 3D plane. Once the FP labels were imported in the matching plane of the CCF on Adobe Illustrator, we used linear translation to stretch the FP labels to fit the CCF roughly. Then, we performed finer alignment manually based on specific landmarks of the brain with distinct contrast (e.g., fiber tracts). We used shade from levels of background autofluorescence and texture from fine myelinated tracks in the Allen CCF to determine anatomical borders. The shade feature was useful for delineating subregions in the isocortex, the hippocampus, the hindbrain, and the cerebellum; the texture feature was useful in the ventral striatum and the medulla. One or two slides before and after in each 2D section were used to ensure the contiguity of 3D labels. In selected areas (e.g., hypothalamus), boundaries were removed entirely and re-drawn based on key features of the CCF and distinct cell populations. In caudal areas, we often used 2–3 different FP planes to create hybrid labels to fit the CCF background as well as cell type specific features of the selected plane. The primary alignment of each label was performed by U.C., followed by a second and independent inspection by Y.K.