Skip to content
Pengcheng Zhou edited this page Sep 27, 2017 · 5 revisions

Welcome to the CNMF_E wiki!

Introduction

CNMF-E is an automated toolbox for detecting neurons and extract their temporal components from calcium imaging data. It's based on the Constrained Nonnegative Matrix Factorization (CNMF) framework, which was originally developed by Eftychios A. Pnevmatikakis and Liam Paninski [1].

The letter 'E' in CNMF-E has two meanings: 1. endoscope; 2. extension. At the beginning, we want to modify the vanilla CNMF to processing micro-endoscopic data, which suffers from large contaminating background fluctuations. The related work was reported in our arXiv preprint [2]. The project was led by Pengcheng Zhou and Liam Paninski. This CNMF-E package is the implementation of our work, which is also built onto the original CNMF.

Our contributions to CNMF implementations are two folds:

  1. a general background model to accurately model the background fluctuations and efficiently separate them from the neural signals;

  2. plenty of functions that make the calcium imaging analysis easy to run and scalable to large-scale datasets.

Highlights

  1. It uses class implementation to easily run the analysis with the least arguments passing.

  2. It is scalable to large-scale datasets easily. It has two modes for dividing the large problem into small problems: patch mode (divide FOV into small patches) and batch mode (segment a long recording into small batches temporally). These two modes can be combined together. This implementation has three important advantages:

  • analyze large data with a common desktop computer.
  • boost the speed of analysis using parallel processing if the computer has a large RAM and multi-core CPU.
  • track the same neuron across multiple days by letting neuron shapes shared across all batch (each session is one batch).
  1. It has a logging system to keep track of the whole processing of the data, which is good for reproducible research.

  2. All subproblem in the model are modularized and users can easily customize the implementation for their distinct problems.

References

[1] Pnevmatikakis, E.A., Soudry, D., Gao, Y., Machado, T.A., Merel, J., Pfau, D., Reardon, T., Mu, Y., Lacefield, C., Yang, W. and Ahrens, M., 2016. Simultaneous denoising, deconvolution, and demixing of calcium imaging data. Neuron, 89(2), pp.285-299. [2] Zhou, P., Resendez, S.L., Stuber, G.D., Kass, R.E. and Paninski, L., 2016. Efficient and accurate extraction of in vivo calcium signals from microendoscopic video data. arXiv preprint arXiv:1605.07266.

Clone this wiki locally