All Tools
Neuroanatomy · Atlas Registration

NeuroFlow

A self-contained, browser-based workflow that takes mouse brain sections from raw histology images to region-resolved counts, registering each section to the Allen Mouse Brain atlas and quantifying labeled signal per brain region, entirely on your own computer.

Launch NeuroFlow Read the paper Runs 100% locally: no upload, no server
Aarushi Rao
Developed by Aarushi Rao, with Henry Oo, Dr. Can Tao & Dr. Guang-Wei Zhang. See the team page →

Video tutorial

A complete walkthrough of the pipeline, from loading raw section images through registration, signal detection, and regional quantification. Follow along in a second tab with your own data.

What it does

Mapping histological sections onto a common reference atlas is an essential but usually fiddly step in neuroscience: existing workflows are powerful, yet an analysis often spans several programs, operating-system-specific components, or specialized software dependencies. NeuroFlow brings the whole pipeline into a single web page that runs locally in your browser, and your images are never uploaded and there is no backend server.

It integrates section cropping, anterior–posterior ordering and coarse atlas-plane assignment, real-time 3D oblique atlas reslicing, affine and nonlinear registration, selected-color signal detection, atlas-based regional quantification, series-level aggregation, and interactive 3D inspection, all as one coherent, seven-stage pipeline. Major controls and intermediate results stay inspectable throughout, a project file preserves your section, registration, mask, and result state, and every stage can be exported to a spreadsheet.

Overview of the NeuroFlow six-stage pipeline
The pipeline at a glance. The interface stages (Cropper, Arranger, Scaler, Quantifier, Analyzer, and Visualizer) carry a section series from preparation through registration, signal detection, regional aggregation, and 3D inspection. Figure from the preprint; the current build adds a dedicated 3D Orienter stage between Arranger and Scaler.

The seven-stage pipeline

1

Cropper · optional

Turn a whole-slide or multi-section source image into clean, editable per-section crops. Draw, move, and resize regions of interest; preview black/white and gamma; and export the prepared sections, each keeping a stable identity so it can be tracked through the rest of the pipeline.

Cropper and Arranger stages
Cropper & Arranger. Source images become editable crops that preserve section identity (A); sections are oriented and ordered anterior→posterior (B); and each is matched to the closest plate from a 24-plate Allen CCFv3 reference grid as a coarse starting point (C).
2

Arranger

Orient each section (rotate 90°, flip horizontally or vertically) and drag them directly into anterior-to-posterior order. Center midlines and exclude or pin sections as needed. Section identity is preserved so you can inspect the series before any registration.

3

3D Orienter · oblique planes

Set each section's true 3D cutting plane before any scaling. Sections matched to an atlas plate are sent here, where you tilt the plane directly (drag to tilt, shift-drag to resize), adjusting anterior–posterior depth, pitch, and yaw against a real-time oblique reslice of the atlas. Sections cut off-axis get an atlas plane that actually matches them, rather than being forced onto the nearest coronal plate.

4

Scaler · registration

Register each section to the Allen Mouse Brain CCFv3, working against the oriented plane from the previous stage. Refine with affine transforms (translate, rotate, scale) and then a nonlinear control-point warp for local alignment. Each section's transform is saved.

Scaler registration stage
Scaler. From an intentionally mismatched coarse plate (A) to a matched oblique atlas reslice (B), then affine (C) and nonlinear control-point (D) refinement, ending in a final overlay aligned to the tissue boundary with the transform saved (E).
5

Quantifier · signal detection

Detect labeled signal by color. Pick a representative signal color by typing it or sampling with the eyedropper, restrict analysis to valid tissue with brush/lasso/erase mask tools, then set a contrast threshold (for green, the score is G − max(R, B)) and a connected-component size cutoff. A cyan preview shows every threshold-positive pixel. NeuroFlow works from source-image pixels only. It does not invent cells or synthetic centroids.

Quantifier signal-pixel detection stage
Quantifier. Choose the signal color (A), define the tissue mask (B), apply a contrast threshold (C) and a component-size cutoff (D), and review the detected pixels in a cyan preview (E), with every parameter inspectable.
6

Analyzer · regional quantification

Turn detected pixels into region-resolved numbers. Each qualifying pixel inherits the CCFv3 region label at its registered coordinate, producing a ranked table of regions (acronym, pixel count, density), a two-level sunburst of major divisions and their constituent regions, and an Allen region-hierarchy tree. Results aggregate across the whole section series.

Analyzer regional quantification stage
Analyzer. Qualifying pixels (A–B) are assigned CCFv3 regions at their registered coordinates (C); a sunburst summarizes major divisions and their regions (D), and a root-to-region hierarchy lists the leading contributions (E), colored to match the Allen atlas.
7

Visualizer · 3D inspection

Inspect the result in 3D. The registered section appears as a clipped plane sitting at its true intersection with the actual Allen CCFv3 whole-brain surface, with the measured signal pixels shown in place and a linked 2D view for full context. Rotate, pan, zoom, and hover to explore.

Visualizer 3D inspection stage
Visualizer. The registered section (A) and its measured signal pixels (B) are shown on the true 3D intersection with the CCFv3 whole-brain surface, alongside a linked 2D section for context (C).

Highlights

Runs locally: no upload, no server Allen Mouse Brain CCFv3 Real-time 3D oblique reslicing Affine + nonlinear registration Region-resolved quantification Project file preserves state

Quick start

Click Launch NeuroFlow, load your section images, and follow the stages left to right: Cropper → Arranger → 3D Orienter → Scaler → Quantifier → Analyzer → Visualizer. Everything runs in your browser and your images never leave your computer.

Reference. Rao A, Oo H, Tao C, Zhang G-W. NeuroFlow: An Integrated, Cross-Platform Workflow for Mouse Brain Atlas Registration and Quantification. bioRxiv, 2026. Figures adapted from the manuscript.   Read on bioRxiv  ·  PDF