search

cellarium-ai / CellMincer

CellMincer is a software package for self-supervised denoising of voltage imaging datasets
BSD 3-Clause "New" or "Revised" License
2 stars 0 forks source link

readme

CellMincer

CellMincer is a self-supervised machine learning framework for voltage imaging denoising models. A visual comparison of voltage imaging data before and after CellMincer is shown below:

Alt Text

An overview of voltage imaging data generation and CellMincer denoising model is shown below:

Alt Text

(a) A simplified schematic diagram of a typical optical voltage imaging experiment (left). The spatially resolved fluorescence response is recorded over time to produce a voltage imaging movie. A key component of CellMincer's preprocessing pipeline is the computation of spatial summary statistics and various auto-correlations from the entire recording, which are concatenated into a stack of global features (right). (b) An overview of CellMincer's deep learning architecture. (c) The conditional U-Net convolutional neural network (CNN). At each step in the contracting path, the precomputed global feature stack is spatially downsampled in parallel ($\mathscr{F}\rightarrow \mathscr{F}'\rightarrow \mathscr{F}''$) and concatenated to the intermediate spatial feature maps. (d) The temporal post-processor neural network. The sequence of pixel embeddings are convolved with a 1D kernel along the time dimension, producing a single vector of length $C$. A multilayer perceptron subsequently reduces this vector to a single value. (e) A comparison of model performance on simulated data before and after introducing global features as a U-Net conditioner. The distributions of PSNR gain are binned by stimulation amplitude. Using global features confers an average increase of 5 dB to the denoiser, roughly corresponding to a 3-fold noise reduction.

Documentation

Our documentation is hosted on ReadtheDocs and includes an overview of our tool's intended use-cases, a walkthrough of our pipeline, detailed explanations of our configuration options, and pretrained models to try CellMincer on your data.

Data availability

Raw and denoised voltage imaging datasets, as well as pretrained models and example configurations, can be found at this Google bucket: gs://broad-dsp-cellmincer-data (refer to downloading from Google Cloud storage).

Preprint and citation

The bioRxiv preprint for CellMincer can be found here. The BibTeX citation:

@article {Wang2024.04.12.589298,
    author = {Brice Wang, Tianle Ma, Theresa Chen, Trinh Nguyen, Ethan Crouse, Stephen J Fleming, Alison S Walker, Vera Valakh, Ralda Nehme, Evan W Miller, Samouil L Farhi, and Mehrtash Babadi},
    title = {Robust self-supervised denoising of voltage imaging data using CellMincer},
    elocation-id = {2024.04.12.589298},
    year = {2024},
    doi = {10.1101/2024.04.12.589298},
    URL = {https://www.biorxiv.org/content/early/2024/04/15/2024.04.12.589298},
    eprint = {https://www.biorxiv.org/content/early/2024/04/15/2024.04.12.589298.full.pdf},
    journal = {bioRxiv}
}

Related Github repositories

CellMincerPaperAnalysis contains notebooks for reproducing the analysis and figures in the preprint.

Optosynth is a voltage imaging simulation framework which generates synthetic data used to optimize and benchmark CellMincer.