Brain Atlas Registration
at Cohort Scale.
Map every coronal section to Allen CCFv3, Waxholm Rat, or your own custom atlas. Affine plus nonlinear warp, per-region quantification, and a headless CLI that processes 100+ brains overnight.
A registration engine built for whole-cohort neuroanatomy.
Four capabilities that turn atlas registration from a per-section chore into a one-command pipeline you can rerun any time.
Affine + Nonlinear Warp
Two-stage registration combines rigid plus affine alignment with B-spline nonlinear warping. Handles tissue distortion, sectioning artifacts, and individual anatomical variation.
ANTs SyN under the hood, with sensible defaults tuned for fixed mouse and rat brain sections. Per-section transform fields are stored and reusable across runs.
Multi-Channel Registration
Register a single reference channel — DAPI, autofluorescence, or Nissl — once, then propagate the transform to every other channel. Quantitative intensities stay intact across markers.
Bidirectional transforms supported: pull atlas labels into image space, or push image features into atlas space for cross-subject comparison.
Per-Region Quantification
Cell counts, fluorescence intensities, and densities tabulated per atlas region. CSV export carries the full ontology hierarchy — pool by parent structure or compare leaf level.
Joins cleanly to ConductGraph publication metadata so cohort results land in a SciencePin-ready manifest without manual reshaping.
Headless Cohort Pipeline
A CLI registers 100+ brains overnight on a single workstation. Per-section QC reports flag anything that needs a human eye, and intermediate transforms cache for restartable runs.
Drop a YAML cohort spec onto a folder of OME-TIFF stacks; results land as per-brain CSVs plus a roll-up Parquet ready for Polars or DuckDB.
One pipeline. Three backends. Your choice of reference frame.
Allen Mouse Brain CCFv3 and Waxholm Sprague Dawley Rat ship out of the box. Bring a custom atlas — primate, neonate, disease model — and the warp engine treats it the same.
Allen CCFv3Mouse · 2017 | Waxholm Rat v4Sprague Dawley · 2019 | Custom AtlasBYO · validated | |
|---|---|---|---|
| Species | Mus musculus | Rattus norvegicus | Any |
| Reference | 1,675 STPT brains | MRI + DTI | Bring your own |
| Resolution | 10 / 25 / 50 µm | 39 µm | Configurable |
| Regions | ~1,300 | 222 | User defined |
| Modalities | Autofluor · Nissl · DAPI | MRI · DTI · histology | Any 2D / 3D channel |
| License | CC BY 4.0 (Allen) | CC BY-SA 4.0 (NITRC) | Yours |
Four steps. One config file. The same pipeline every time.
The same calls run in the GUI, in a Jupyter notebook, or in the headless CLI — so what you validate on one brain runs unchanged on the cohort.
Load
Import preprocessed sections from ConductPrep or any OME-TIFF source. Auto-detect orientation and approximate bregma.
register.load('cohort.yaml')Align
Coarse rigid plus affine alignment to the atlas template via the reference channel. Manual landmarks available for atypical sections.
.align(method='affine')Warp
Nonlinear deformation field maps every pixel to atlas coordinates. The inverse transform pulls atlas labels back into your image space.
.warp(grid='bspline')Quantify
Per-region cell counts, intensities, and densities. Export CSV with full ontology hierarchy for downstream analysis in R, Python, or Prism.
.export('cohort.csv')Built on the methods that the field already cites.
The atlas backends, transform model, and quantification reporting all map to peer-reviewed standards — so reviewers see a methods section they recognise.
A Common Coordinate Framework for the Mouse Brain.
Waxholm Space atlas of the rat brain hippocampal region: Three-dimensional delineations.
Symmetric diffeomorphic image registration with cross-correlation: evaluating automated labeling of elderly and neurodegenerative brain.
Stop drawing ROIs by hand.
Map every section to atlas coordinates and let the data speak by region. Demo on your own brains, or send us a folder and we will run the cohort.
