This repository contains a multistep algorithm to detect swelling of mini-organs (organoids) in subsequent microscopy images (movie). All notebooks are specifically written for deployement using Google Colab.
First, organoids are detected using the organoid_recognition.ipynb notebook. In this notebook, organoids are detected for each frame of the microscopy movie. A pre-trained neural network OrgaQuant, described in this open-access paper, is used for organoid detection.
| Microscopy Image | Detected Organoids |
|---|---|
 |
 |
In a second post_processing.ipynb, we estimate the average swell rate of the detected organoids in the movie. First, we use particle tracking to track the location of each organoids . Next, we use linear regression to quantify swelling of each organoid. Using the swell rate of each organoid, we can estimate the average swell rate of all organoids in a movie.
| Particle Tracking | Swell regression | Swell Rates |
|---|---|---|
 |
 |
 |
| x, y position for each organoid in all 13 movie frames | Linear regression to estimate the swell rate of all detected organoids in the movie | The mean swell rate of multiple movies. No organoid swelling is observed for most wells, however, clear swelling of organoids is observed in well 16. |