• Corpus ID: 2695928

Fast Constrained Non-negative Matrix Factorization for Whole-Brain Calcium Imaging Data

@inproceedings{Friedrich2016FastCN,
  title={Fast Constrained Non-negative Matrix Factorization for Whole-Brain Calcium Imaging Data},
  author={Johannes Friedrich and Daniel Soudry and Yu Mu and Jeremy Freeman and Misha B. Ahrens and Liam Paninski},
  year={2016}
}
Advances in optical recording technologies allow whole-brain recordings with single cell resolution of small animals, such as larval zebrafish. A crucial step for further neural analysis is the move from voxel-space to neural-space, i.e. the detection of all neurons and extraction/demixing of how each neurons activity evolves in time. The sheer amount of data has led experimenters towards simple but fast methods such as averaging voxels in a small neighborhood around identified neuron centers… 

Figures from this paper

SCALPEL: EXTRACTING NEURONS FROM CALCIUM IMAGING DATA.

This paper considers the first step in the analysis of calcium imaging data-namely, identifying the neurons in a calcium imaging video, and proposes a dictionary learning approach for this task, which is implemented in the R package scalpel.

Multi-scale approaches for high-speed imaging and analysis of large neural populations

A demixing algorithm can be applied to accurately recover each neuron’s activity from the low-resolution data by exploiting the high-resolution cellular maps estimated in the first imaging phase, and by decreasing the spatial resolution in the second phase, this approach can compress the video data significantly, and potentially acquire images over an order-of-magnitude larger area, or image at significantly higher temporal resolution.

Computational Methods for Neuron Segmentation in Two-Photon Calcium Imaging Data: A Survey

An overview of the main algorithms taxonomized in three categories (signal processing, matrix factorization and machine learning-based approaches) are provided, highlighting the main advantages and disadvantages of each category.

VolPy: Automated and scalable analysis pipelines for voltage imaging datasets

VolPy is presented, the first turn-key, automated and scalable pipeline to pre-process voltage imaging datasets that features fast motion correction, memory mapping, segmentation, and spike inference, all built on a highly parallelized and computationally efficient framework that optimizes memory and speed.

Submitted to the Annals of Applied Statistics SCALPEL : EXTRACTING NEURONS FROM CALCIUM IMAGING DATA By

A dictionary learning approach is proposed to develop a dictionary containing a huge number of candidate neurons and apply the dictionary to select neurons and estimate their corresponding activity over time, using a sparse group lasso optimization problem.

Auto-Encoding Topographic Factors

Self-Encoding Topographic Factors is presented, a novel variational inference scheme to decompose irregular observations on a lattice into a superposition of low rank sources that gives an implicit mapping to a unique latent representation while simultaneously forcing the posterior to model group variability in spatial structure.

Exponential Family Embeddings

On all three applications—neural activity of zebrafish, users' shopping behavior, and movie ratings—the exponential family embedding models are found to be more effective than other types of dimension reduction and better reconstruct held-out data and find interesting qualitative structure.

Data-Adaptive Modeling using Convex Regression

This work presents a method that adaptively selects covariates to include in the model and for those included, models their conditional associations with the outcome as piecewise constant functions with adaptivelychosen knots.

Efficient and accurate extraction of in vivo calcium signals from microendoscopic video data

A new constrained matrix factorization approach to accurately separate the background and then demix and denoise the neuronal signals of interest is described, which substantially improved the quality of extracted cellular signals and detected more well-isolated neural signals, especially in noisy data regimes.

References

SHOWING 1-7 OF 7 REFERENCES

Simultaneous Denoising, Deconvolution, and Demixing of Calcium Imaging Data

Detecting cells using non-negative matrix factorization on calcium imaging data

Mapping brain activity at scale with cluster computing

A library of analytical tools called Thunder built on the open-source Apache Spark platform for large-scale distributed computing that implements a variety of univariate and multivariate analyses with a modular, extendable structure well-suited to interactive exploration and analysis development.

Whole-brain functional imaging at cellular resolution using light-sheet microscopy

Light-sheet microscopy is used to record activity from the entire volume of the brain of the larval zebrafish in vivo at 0.8 Hz, capturing more than 80% of all neurons at single-cell resolution, demonstrating how this technique can be used to reveal functionally defined circuits across the brain.

Hierarchical ALS Algorithms for Nonnegative Matrix and 3D Tensor Factorization

This paper proposes to use local cost functions whose simultaneous or sequential (one by one) minimization leads to a very simple ALS algorithm which works under some sparsity constraints both for an under-determined and overdetermined model.

Scikit-learn: Machine Learning in Python

Scikit-learn is a Python module integrating a wide range of state-of-the-art machine learning algorithms for medium-scale supervised and unsupervised problems. This package focuses on bringing