roicat

Welcome to ROICaT

🎉 CONTRIBUTIONS WELCOME! 🎉
See the TODO section

• Documentation: https://roicat.readthedocs.io/en/latest/
• Discussion forum: https://groups.google.com/g/roicat_support
• Technical support: Github Issues

Region Of Interest Classification and Tracking ᗢ

A simple-to-use Python package for automatically classifying images of cells and tracking them across imaging sessions/planes.

Why use ROICaT?

• It's easy to use. You don't need to know how to code. You can use the interactive notebooks or online app to run the pipelines with just a few clicks.
• It's accurate. ROICaT was desgined to be better than existing tools. It is capable of classifying and tracking neuron ROIs at accuracies approaching human performance out of the box.
• It's fast and computational requirements are low. You can run it on a laptop. It was designed to be used with >1M ROIs, and can utilize GPUs to speed things up.

With ROICaT, you can:

• Classify ROIs into different categories (e.g. neurons, dendrites, glia, etc.).
• Track ROIs across imaging sessions/planes (e.g. ROI #1 in session 1 is the same as ROI #7 in session 2).

What data types can ROICaT process?

• ROICaT can accept any imaging data format including: Suite2p, CaImAn, CNMF, NWB, raw/custom ROI data and more. See below for details on how to use any data type with ROICaT.

How to use ROICaT

TRACKING:

• Online App: Good for first time users. Try it out without installing anything.
• Interactive notebook (also available in colab) $~$
• Command line interface script:

roicat --pipeline tracking --path_params /path/to/params.yaml --dir_data /folder/with/data/ --dir_save /folder/save/ --prefix_name_save expName --verbose

CLASSIFICATION:

• Interactive notebook - Drawing. (also available in colab) $~$: $~~$ Use mouse to draw circles around regions of a UMAP to classify ROIs.
• Interactive notebook - Simple interactive Labeling: $~~$ Images of ROIs are displayed and you use the keyboard to manually label them.
• Interactive notebook - Interactive labeling with drawing: $~~$ Draw circles around regions of a UMAP to subselect ROIs for manual labeling.
• Interactive notebook - Train classifier
• Interactive notebook - Inference with classifier

OTHER:

• Custom data importing notebook
• Use the API to integrate ROICaT functions into your own code: Documentation.
• Run the full tracking pipeline using the CLI or roicat.pipelines.pipeline_tracking with default parameters generated from roicat.util.get_default_paramaters() saved as a yaml file.

Installation

ROICaT works on Windows, MacOS, and Linux. If you have any issues during the installation process, please make a github issue with the error.

0. Requirements

• Anaconda or Miniconda.
• If using Windows: Microsoft C++ Build Tools
• The below commands should be run in the terminal (Mac/Linux) or Anaconda Prompt (Windows).

1. (Recommended) Create a new conda environment

conda create -n roicat python=3.12
conda activate roicat

You will need to activate the environment with conda activate roicat each time you want to use ROICaT.

2. Install ROICaT

pip install roicat[all]
pip install git+https://github.com/RichieHakim/roiextractors

Note on zsh: if you are using a zsh terminal, change command to: pip3 install --user 'roicat[all]'
Note on installing GPU support on Windows: see GPU Troubleshooting documentation.
Note on opencv: The headless version of opencv is installed by default. If the regular version is already installed, you will need to uninstall it first.

3. Clone the repo to get the notebooks

git clone https://github.com/RichieHakim/ROICaT

Then, navigate to the ROICaT/notebooks/jupyter directory to run the notebooks.

Upgrading versions

There are 2 parts to upgrading ROICaT: the Python package and the repository files which contain the notebooks and scripts.
Activate your environment first, then...
To upgrade the Python package, run:

pip install --upgrade roicat[all]

To upgrade the repository files, navigate your terminal to the ROICaT folder and run:

git pull

General workflow:

• Pass ROIs through ROInet: Images of the ROIs are passed through a neural network which outputs a feature vector for each image describing what the ROI looks like.
• Classification: The feature vectors can then be used to classify ROIs:
  • A simple regression-like classifier can be trained using user-supplied labeled data (e.g. an array of images of ROIs and a corresponding array of labels for each ROI).
  • Alternatively, classification can be done by projecting the feature vectors into a lower-dimensional space using UMAP and then simply circling the region of space to classify the ROIs.
• Tracking: The feature vectors can be combined with information about the position of the ROIs to track the ROIs across imaging sessions/planes.

Run the app locally

Although, we recommend transitioning to using the notebooks or CLI instead of the app, you can download and run the app locally with the following command:

sudo docker run -it -p 7860:7860 --platform=linux/amd64 --shm-size=10g registry.hf.space/richiehakim-roicat-tracking:latest streamlit run app.py

TODO:

algorithmic improvements:

• Add in method to use more similarity metrics for tracking
• Coordinate descent on each similarity metric
• Add F and Fneu to data_roicat, dFoF and trace quality metric functions
• Add in notebook for demonstrating using temporal similarity metrics (SWT on dFoF)
• Make a standard classifier
• Try other clustering methods
• Make image aligner based on image similarity + RANSAC of centroids or s_SF
• Better post-hoc curation metrics and visualizations
• Improve non-rigid image registration methods (border performance)
• Make non-rigid image registration optional

code improvements:

• Finish ROIextractors integration
• Update automatic regression module (make new repo for it)
• Switch to ONNX for ROInet
• Some more integration tests
• Figure out RNG / OS differences issues for tests
• Add more documentation / tutorials
• Make a GUI
• Add settings to the GUI
• [ ] Make a Docker container
• Make colab demo notebook not require user data
• Make a better CLI
• Switch to pyproject.toml
• Improve params.json / default params system
• Spruce up training code

other:

• Write the paper
• Make tweet about it
• Make a video or two on how to use it
• Maybe use lightthetorch for torch installation
• Better Readme
• More documentation
• Make a regression model for in-plane-ness
• Formalize bounty program