WATERS

Spike detection based on continuous wavelet transform with data-driven templates.

Description

This method is based on:

Nenadic Z, Burdick JW.
Spike detection using the continuous wavelet transform.
IEEE Trans Biomed Eng. 2005;52(1):74-87.
doi:10.1109/TBME.2004.839800

Through adapting custom wavelets based on spike waveforms, it creates a family of templates that are then scaled:

Horizontally in time domain,
Vertically in voltage (amplitude) domain.

This allows for a more robust spike detection that accounts for the physiological spike waveforms (as opposed to abstract wavelets).

Spike detection pipeline

Filter the raw voltage trace (3rd order Butterworth, 600 Hz - 8 kHz).
Using threshold-based method detect n ∈ [50, 1000] spikes.
Using the aggregated median spike waveform adapt a custom wavelet.

Spike Overlay	Average waveform	Adapted wavelet

Run spike detection scaling the custom wavelet across scales.

Compare with threshold-based method and built-in wavelets from MATLAB Wavelet Toolbox.

Installation

Clone git repository:

git clone https://github.com/jeremi-chabros/CWT.git

Requires MATLAB with Wavelet Toolbox and Signal Processing Toolbox.

Usage

Preferred method is executing (or calling from command window) getSpikesApp.m

Toggle between setting path to the folder with data or loading specific file(s)
Select files/folder
Select output folder (by default saves in current directory)
Set parameters and save them
Run spike detection

Tip: Hover mouse over a parameter to display its value and usage.

Organization of different functions is as follows:

├── batchDetectSpikes.m
│   └── detectSpikesCWT.m
│       ├── getTemplate.m
│       │   └── detectSpikesThreshold.m
│       ├── adaptWavelet.m
│       └── detectSpikesWavelet.m

This method is still under development and troubleshooting and hence frequent git pull is recommended:

git stash -A

git pull

License

MIT

Timothysit / WATERS