Thomas Spike Sorter

Simultaneous recordings with multi-channel electrodes, e.g. tetrodes, are increasingly used for studying neuronal information processing. The recorded extracellular signals contain action potentials (spikes) of a number of adjacent or distant neurons. These signals must be sorted correctly into spike trains of individual neurons. However, several mathematical methods have been proposed to solve this task. Spike sorting is difficult in practice, since the extracellular recorded signals are often non-stationary and generally corrupted by biological and electronic noise or artefacts. From a theoretical point of view one has to solve a cocktail party problem with time-varying sources and superimposed interferences. Recently a novel spike sorting algorithm has been developed [1]. To allow a user to interact with this algorithm with images rather than text commands we developed a graphical user interface (GUI).

Key features:

  • Spike detection with non-linear energy operator
  • Semi-automatic clustering by principal component analysis
  • Enhancing of spike sorting by linear filters for SNR improvement
  • Multichannel signal and result visulization
  • Template view for enhanced controlling of sorting results
  • Easy parameter settings
  • Fast and interactive data processing
  • FREE demo version available

With the spike sorting method used in the Thomas Tetrode Spike Sorter, spike shapes are guessed from the signal by spike detection via well-known nonlinear energy-operator and classification via principal component clustering. The information found is used, to build optimal linear filters that improve the SNR for better detection and classification of spikes.

Figure 1: Graphical user interface of the Thomas Tetrode Spike Sorter

The method is optimized for multichannel recording electrodes, like stereotrodes or, especially, tetrodes. It combines temporal and spatial information provided by multichannel recordings.

The separation of overlapping spikes is inherent to this spike sorting method and thus enhances the detection and classification of spikes in recordings with many neurons.

The spike sorting method was originally developed in a cooperation between Thomas RECORDING and the Technical University in Berlin (Prof. K. Obermayer and colleagues). The method has been published by Franke et al. [1].

Together with the research group of Professor Thomas Schanze at the Technical University (THM) in Gießen, Germany we recently improved the algorithm and developed a graphical user interface (GUI) [2]. These improvements assured that the parameterization and handling of the spike sorting tool gets easy. 

This Tetrode Spike Sorter was developed in the Bernstein research project “Memory Network” (Project Partners: Thomas RECORDING GmbH, Max Planck Institute for Brain Research, Technical University Berlin) and supported by grant of the German Ministry of Education and Research (grant numbers 01GQ0741, 01GQ0742, 01GQ0743)

Further developments of the project were made in a ZIM research project “Development of high resolution Multi-Heptode-Recording Systems for Brain Research Applications” (Project partners: Technical University Mittelhessen in Giessen and Thomas RECORDING GmbH) and supported by grant of the German Ministry of Economics and Technology (grant number KF2780402AK2)

Publications

[1] F. Franke, M. Natora, C. Boucsein, M. Munk, and K. Obermayer: An online spike detection and spike classification algorithm capable of instantaneous resolution of overlapping spikes. J. Comput. Neurosci., 127-148, 2009.

[2] Doerr, C., Hoehl, D., Thomas, U., & Schanze, T. (2012). ROC-testing of a spike sorting algorithm. Biomed Tech , 57, 1.  [link]    

[3] Doerr, C., Hoehl, D., Thomas, U., & Schanze, T. (2012). Testing of a Spike Sorting Algorithm and GUI development. Workshop "Innovative Verarbeitung bioelektrischer und biomagnetischer Signale" - bbs2012, 19.-20.04.2012, Berlin.

[4] Doerr, C., & Schanze, T. (2013). Analytic Signal Based Detection of Extracellular Action Potentials. Biomed Tech , 58, 1.  [link]   

[5] Doerr, C., Hoehl, D., Thomas, U., & Schanze, T. (2013a). Testing and Improvement of a Spike Sorting Algorithm. NWG2013 - Tenth Göttingen Meeting of the German Neuroscience Society.   

[6] Doerr, C., Hoehl, D., Thomas, U., & Schanze, T. (2013b). Spike-Sorting of Clinical Tetrode-Recordings. Automed 2013 Workshop, Dresden. 

[7] Doerr, C., Hoehl, D., Thomas, U., & Schanze, T. (2013c). Enhancements of a spike sorting algorithm: GUI development and detection optimization. Bernstein Conference 2013, Tübingen.   [link]

[8] Doerr, C., & Schanze, T. (2014). Spike sorting performance depends on the number of signal channels: a simulation study. Workshop "Innovative Verarbeitung bioelektrischer und biomagnetischer Signale" - bbs2014, 10.-11.04.2014, Berlin.

[9] Doerr, C., & Schanze, T. (2014). The multitrode-effect influences the spike sorting performance: a simulation study. Biomed Tech , 59, 1, accepted

[10] Doerr, C. & Schanze, T. (2015). Are Heptodes better than Tetrodes for spike sorting? ScienceDirect; IFAC-PapersOnLine; Volume 48, Issue 20, 2015, Pages 94-99

[11] Eppinger, R., Schanze, T. (2017): An evaluation of two spike sorting algorithms: Heptode Spike Sorter versus WaveClus. Poster presentation at the 12th Goettingen Meeting of the German Neuroscience Society, March 2017

 

Videos