exo2micro Documentation

exo2micro is an image registration and fluorescence subtraction pipeline for pre/post-stain microscopy. It has been designed to process paired pre-stain and post-stain microscopy images by (a) aligning the images using an iterative process that allows for translation, rotation, shear, and magnification differences between pre- and post- stain images, (b) estimating the autofluorescent mineral background signal, and (c) optimally subtracting pre- from post-stain images to reveal microbe-only signal. The software is still in active development.

This software was developed as a collaboration between the Follette Lab at Amherst College, which specializes in astronomical image processing, especially of extrasolar planets and the Marlow Lab at Boston University, which specializes in microscopy imaging of astrobiologically interesting terrestrial rock samples. Development of this software was supported by Heising-Simons Foundation grant #2022-3992 issued as part of the Scialog: Signatures of Life in the Universe conference series. The software is named for the title of our original grant proposal: “Exoplanets to Microbes: Using Astronomical Image Processing Techniques to Detect Microbes in Astrobiological Contexts”.

The current software was written and is maintained by Kate Follette. Undergraduate researchers Giselle Hoermann, Kinsey Cronin, Jessica Laboissiere, Sarah Vierling, and Suyash Deshmukh contributed significantly to early versions of some of its functionalities.

Note

This documentation is split into two tracks. If you’re here to process images, start with the Users track. If you’re here to extend, script, or tune the pipeline, head to the Developers track. Most people will want the Users track.

Users

• Users Track
  • Before You Start
  • Installation
  • Quickstart
  • GUI Tour
  • Scale Methods
  • Interpreting Results
  • Zoom & Inspect
  • Memory and Performance
  • Troubleshooting

Developers

• Developers Track
  • Conceptual Overview
  • Scripting API
  • Parameter Reference
  • Scale Estimation Methods
  • Extending exo2micro
  • API Reference

Reference

• Migration Guide

What exo2micro does

Take two fluorescence microscopy images of the same mineral sample — one before staining (autofluorescent mineral background only) and one after (background + any microbes that took up the stain) — and produce a difference image in which microbial signal dominates.

The challenge is that the two images aren’t perfectly aligned (the sample may have shifted, rotated, or been imaged at a slightly different magnification) and the autofluorescent background has a slightly different overall brightness in the two images (because autofluorescence itself can vary with staining chemistry and sample orientation). exo2micro solves both: it registers the pre-stain image to the post-stain image using a multi-stage alignment pipeline, then estimates and subtracts a scaled version of the background.

Four stages, roughly:

1. Padding — load raw TIFFs onto a common zero-padded canvas.

2. Boundary alignment — phase correlation + ICP on the sample outline.

3. Interior alignment — SIFT feature matching on interior structure.

4. Diagnostics — estimate the background scale factor (via a Moffat fit on the log-ratio distribution) and produce the scaled difference image, plus diagnostic plots.

Quick start

The fastest way in, for most people, is the interactive GUI:

from exo2micro.gui import launch
launch()

For scripting:

import exo2micro as e2m
run = e2m.SampleDye('CD070', 'SybrGld_microbe')
run.run()

For everything else, pick a track above.

Version

This documentation is for exo2micro 2.3. The full release history lives in CHANGELOG.md at the repository root. See Migration Guide for upgrading from earlier versions.