TU Berlin

Neural Information ProcessingFunctional Imaging Methods: Source Separation Techniques

Neuronale Informationsverarbeitung

Page Content

to Navigation

Functional Imaging Methods: Source Separation Techniques

Lupe

Optical recording techniques have become a widespread technique for measuring the activity of neural populations. Typically, the optical signal is generated by a mixture of sources, where only some of them are related to the optical signal which is correlated with the neural activity of interest. In addition, the signal-to-noise ratio is often very low, in particular, for single condition maps or and / short recording sequences. In this project we investigated whether recently blind source separation techniques (ICA, extended spatial decorrelation, generalized linear models, etc.) can be properly extended to cope with abovementioned challenges. For optical imaging of intrinsic signals we found, that second order methods based on spatial decorrelation algorithms provided the best results. Methods were applied to optical recording data using intrinsic signals as well as to calcium imaging data.

Acknowledgement: Research was funded by DFG, Wellcome Trust, and Technische Universität Berlin.

Selected Publications:

Principal Component Analysis and Blind Separation of Sources for Optical Imaging of Intrinsic Signals
Citation key Stetter2000a
Author Stetter, M. and Schießl, I. and Otto, T. and Sengpiel, F. and Hübener, M. and Bonhoeffer, T. and Obermayer, K.
Pages 482 – 490
Year 2000
DOI 10.1006/nimg.2000.0551
Journal Neuroimage
Volume 11
Publisher Elsevier
Abstract The analysis of data sets from optical imaging of intrinsic signals requires the separation of signals, which accurately reflect stimulated neuronal activity (mapping signal), from signals related to background activity. Here we show that blind separation of sources by Extended Spatial Decorrelation (ESD) is a powerful method for the extraction of the mapping signal from the total recorded signal. ESD is based on the assumptions, $(i)$ that each signal component varies smoothly across space and $(ii)$ that every component has zero cross-correlation functions with the other components. In contrast to the standard analysis of optical imaging data, the proposed method $(i)$ is applicable to non-orthogonal stimulus-conditions, $(ii)$ can remove the global signal, blood-vessel patterns and movement artifacts, $(iii)$ works without ad hoc assumptions about the data structure in the frequency domain, and $(iv)$ provides a confidence measure for the signals (Z-score). We first demonstrate on orientation maps from cat and ferret visual cortex, that Principal Component Analysis (PCA), which acts as a preprocessing step to ESD, can already remove global signals from image stacks, as long as data stacks for at least two – not necessarily orthogonal – stimulus conditions are available. We then show that the full ESD analysis can further reduce global signal components and – finally – concentrate the mapping signal within a single component both for differential image stacks and for image stacks recorded during presentation of a single stimulus.
Bibtex Type of Publication Selected:sources
Link to publication Link to original publication Download Bibtex entry

Navigation

Quick Access

Schnellnavigation zur Seite über Nummerneingabe