DER algorithm

Duplicate Event Removal algorithm - Artifact detection in human single unit recordings

Readme for the Duplicate Event Removal algorithm:
https://github.com/Geaht/DER

Version 1.0, released February 2021

This project is licensed under the terms of the Mozilla Public License Version 2.0.
Copyright (c) by Gert Dehnen, Marcel S. Kehl, Florian Mormann and the University of Bonn Medical Center
This software was tested with MATLAB (R2018a) on Linux, Windows and MacOS X

---

The DER algorithm was written in MATLAB2018a. To run the code the following MATLAB packages are required:

• Statistics and Machine Learning Toolbox
• Wavelet Toolbox

How to use the code

Download the source code from this repository and add it to your MATLAB path. To run the automated artefact detection you need to execute DER.m in MATLAB.
Please provide the data path and define the spike-sorting algorithm used ('Wave_clus' or 'Combinato') as:

DER(data_path,'Combinato')

Structure of the code

The detection pipeline is structured in three parts:

• Part I - Detection of artifacts within different wire bundles
  The detection of artifacts across different bundles is done by der_detectArtifacts.m.

• Part II - Detection of spikes within channels of same wire bundle
  The detection of artifacts within bundles is done by der_detectDuplicateSpikes.m.

• Part III - Detection of suspicious cross-correlations
  Calculation of all cross-correlations is performed by der_cal_spike_cross_corr_mat.m. Spike events in suspicious central bins of the cross-correlograms are identified by der_detect_cross_corr_spikes.m.

Data structures

Within this pipeline the data structure needed for detection is created by der_get_spikeInfos.m. This script generates a MATLAB table containing all required information on the recorded spike events. These structure includes the timestamp and the waveform of the spike event (64 data points). Additionally, information on the recorded clusters (cluster ID, unit class, channel number, detection threshold, wire bundle ID, anatomical region) as well as a detection label is stored.

The DER alogorithm is currently compatible with the output structures of Combinato Spike Sorting and Wave_clus. In order to apply the algorithm to other spike sorting algorithms you can easily adjust der_get_spikeInfos.m and der_save_spikeinfos.m to your data structure.

Detected spike events are labeled individually for each part of the detection pipeline using prime factor labels. The detection label is given by the procuct of the following prime factors:

Factor	Part detected
2	across bundles (Part I)
3	within the same channel (Part II)
5	within the same bundle (but different channel) (Part II)
7	based on the cross-correlograms (Part III)

References

Cite as:

Dehnen, G.; Kehl, M.S.; Darcher, A.; Müller, T.T.; Macke, J.H.; Borger, V.; Surges, R.; Mormann, F. Duplicate Detection of Spike Events: A Relevant Problem in Human Single-Unit Recordings. Brain Sci. 2021, 11, 761. https://doi.org/10.3390/brainsci11060761