Extracting spatial-temporal coherent patterns in large-scale neural recordings using dynamic mode decomposition.

By Bingni W. Brunton, Lise A. Johnson, Jeffrey G. Ojemann, J. Nathan Kutz. Journal of Neuroscience Methods, 2014

486 citations recorded by the source corpus.

BACKGROUND There is a broad need in neuroscience to understand and visualize large-scale recordings of neural activity, big data acquired by tens or hundreds of electrodes recording dynamic brain activity over minutes to hours. Such datasets are characterized by coherent patterns across both space and time, yet existing computational methods are typically restricted to analysis either in space or in time separately. NEW METHOD Here we report the adaptation of dynamic mode decomposition (DMD), an algorithm originally developed for studying fluid physics, to large-scale neural recordings. DMD is a modal decomposition algorithm that describes high-dimensional dynamic data using coupled spatial-temporal modes. The algorithm is robust to variations in noise and subsampling rate; it scales easily to very large numbers of simultaneously acquired measurements. RESULTS We first validate the DMD approach on sub-dural electrode array recordings from human subjects performing a known motor task. Next, we combine DMD with unsupervised clustering, developing a novel method to extract spindle networks during sleep. We uncovered several distinct sleep spindle networks identifiable by their stereotypical cortical distribution patterns, frequency, and duration. COMPARISON WITH EXISTING METHODS DMD is closely related to principal components analysis (PCA) and discrete Fourier transform (DFT). We may think of DMD as a rotation of the low-dimensional PCA space such that each basis vector has coherent dynamics. CONCLUSIONS The resulting analysis combines key features of performing PCA in space and power spectral analysis in time, making it particularly suitable for analyzing large-scale neural recordings.

Subfields: Cognitive Neuroscience

View Paper: https://www.semanticscholar.org/paper/b08b71828179ebd541b9da3c5e068406fc3f2070

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Cognitive Neuroscience

Citations in the corpus

Learning Koopman Invariant Subspaces for Dynamic Mode Decomposition (2017, background): “, [6]–[9]) and also for engineering tasks, such as signal processing and machine learning.”
Data-driven spectral analysis of the Koopman operator (2017, background): “Recent applications include fluid dynamics [36, 2], power grids [32], neurodynamics [5], energy efficiency [12], molecular physics [48] and data fusion [47].”
Machine Learning Control – Taming Nonlinear Dynamics and Turbulence (2016, methodology): “The DMD method has been applied to numerous problems beyond fluid dynamics [237, 228], where it originated, including epidemiology [217], video processing [124, 99], robotics [25], and neuroscience [40].”
Ergodic Theory, Dynamic Mode Decomposition, and Computation of Spectral Properties of the Koopman Operator (2016, methodology): “Using this framework, they were able to extract a linear system with intermittent forcing that could be used for local predictions of chaotic systems (also see [4] and [2]).”
Sparse Identification of Nonlinear Dynamics with Control (SINDYc) (2016, methodology): “In a related modeling framework, we have developed an algorithm for the sparse identification of nonlinear dynamics (SINDY) from data (Brunton et al., 2016b), relying on the fact that most dynamical systems of interest have relatively few nonlinear terms in the dynamics out of the family of…”
Data-driven discovery of Koopman eigenfunctions for control (2017, unknown): “In addition to fluid dynamics, DMD has been widely applied to a range of problems in neuroscience [18], robotics [14], epidemiology [94], and video processing [40, 44, 70].”
Generalizing Koopman Theory to Allow for Inputs and Control (2016, methodology): “In a very short period of time since, DMD theory has been applied with great success to a broad set of domain sciences including complex fluid flows [45, 43, 44, 13, 2, 48, 47], foreground/background separation in video streams [14], epidemiology [37], and neuroscience [3].”
Multiresolution Dynamic Mode Decomposition (2015, background): “Fields like neuroscience, which are rich in multi-scale, complex dynamics are also ideal candidates for exploration using the mrDMD infrastructure as DMD has already had recent demonstrated success in this arena [23].”
On Convergence of Extended Dynamic Mode Decomposition to the Koopman Operator (2017, background): “This has propelled the theory to wide use on a diverse set of applications such as fluid dynamics [23, 10], power grid dynamics [18], neurodynamics [2], energy efficiency [7], molecular dynamics [29] and data fusion [27].”
Modal Analysis of Fluid Flows: An Overview (2017, methodology): “Moreover, as DMD is purely a data-driven algorithm without the requirement for governing equations, it has been widely applied beyond fluid dynamics: in finance [92], video processing [93– 95], epidemiology [96], robotics [97], and neuroscience [98].”
Computation Through Neural Population Dynamics (2020, background): “(d) Two different neural trajectories (magenta and purple curves) can result from following the neural population dynamics from different initial conditions (magenta and purple squares). reduction techniques (e.g., Cunningham & Yu 2014, Brunton et al. 2016, Kobak et al. 2016, Williams et al. 2018).”
Koopman Invariant Subspaces and Finite Linear Representations of Nonlinear Dynamical Systems for Control (2015, background): “DMD has been broadly applied to a wide range of applied domains, including fluid dynamics [10, 18–21], neuroscience [22], robotics [23], epidemiology [24], and video processing [25–27].”