28 Oct 2026 · 2 min read
Getting started with AxoDen: make your first analysis reproducible
A practical checklist for your first AxoDen analysis: choose an example, define the measurement, inspect the signal and record a reproducible workflow.
AxoDen is an open-source tool for quantifying labelled axonal density in microscopy images. Before applying it to a new experiment, use a public example to understand the workflow and decide which choices you need to record.
This walkthrough is a starting checklist, rather than a claim that one setting fits every microscopy dataset. The repository, API documentation and paper provide the implementation details.
1. Start with an example, not your full experiment
Download the public example images from the repository’s test_images folder. Keep an untouched copy and record the repository commit or release you used. Start with one image so you can inspect each output before moving to a batch.
Choose the interface that fits your work: the web app for exploring the workflow, the desktop interface for local use, or the Python package for integration with an analysis pipeline. Follow the installation and input requirements in the README for the version you are using.
2. Write down the measurement
Before running the analysis, define your region of interest and what the reported density will represent. Check that the input image, dimensions, channel and region selection match those assumptions. Avoid treating output from differently prepared images as directly comparable without checking the acquisition and processing choices.
3. Inspect what the method calls signal
AxoDen uses Otsu thresholding. Inspect the resulting separation of signal and background and compare it with the original image. Look for obvious background being counted as signal, labelled structures being excluded, and regions that do not correspond to the intended anatomy.
If the separation looks unsuitable, investigate image preparation and the suitability of the method before interpreting the number. A successful software run is different from a justified scientific measurement.
4. Save a small analysis record
| Record | Why it matters |
|---|---|
| Software release or commit | Identifies the implementation you used. |
| Original image and identifier | Connects the result to the input. |
| Image preparation and region choices | Makes the measurement definition inspectable. |
| Settings, outputs and quality notes | Supports rerunning and reviewing the analysis. |
5. Test reproducibility before scaling up
Run the same example again with the same inputs and settings. Check that the outputs agree. Then try a small set of images representative of your own experiment and review the signal separation on each. Decide how you will handle unsuitable images before analysing the complete dataset.
The resulting density measure is not, by itself, an axon count or a measure of functional connectivity. Its interpretation depends on the biological question and the acquisition and analysis design.
Explore the AxoDen project or get in touch about quantitative image analysis.