Nev2lkit enhanced

Nev2lkit enhanced: An enhanced version of nev2lkit

Nev2lkit enhanced can sort successfully larger amount of spikes and adapt to low SNRs.

Nev2lkit is a preprocessor for the analysis of intra- or extracellular neuronal recordings. It can plot the data of a 'format-supported' opened file and automatically sort the units, a procedure that is reffered to as Spike Sorting.

Nev2lkit was developed by Daniel Micol for the Unit of Neuroprosthesis and Visual Rehabilitation, Institute of Bioengineering, University Miguel Hernandez of Elche under Cortivis project.

We chose to partially alter nev2lkit's source code, in order to provide enhanced functionality to the spike sorting process. The following features were added, enhancing nev2lkit's capabilities.

1. Variable Length Data Window
The ability of applying Principal Component Analysis on a variable-length data window was implemented, following 0.1 ms accuracy. The data window defines the spike waveform length that is taken into account in the spike extraction process and afterwards inserted into PCA. With this feature, the software can adopt in variable-length spiking activity and avoid the insertion of artifacts.

2. Variability of Principal Components
The ability to select the number of principal components used to perform spike waveform representation by PCA, was also added. The number of engaging principal components (between 2 & 7) can now be selected.

3. Calculation and Display of Variance
In spike waveform representation by the principal components, the percentage variance captured by that subspace is computed and displayed in the program's graphical interface.

4. Spike Count in Each Cluster
Besides the total count of the detected spikes by the extraction process, the ability of displaying the number of spikes assigned to each cluster (as an additional output of the clustering process), was added.

These additional features are demonstrated in action below:

An extracellular recording capturing spontaneous activity of the beetle Tenebrio Molitor respiratory motoneurons, recorded with “hook” electrodes in vitro from the right nerve of its 3rd abdominal ganglion as described in (1), was opened by the program. The time scale was initially set at 2 ms. In order to adopt to variable-length spiking activity we 'tuned' the time-window to 2.6 ms, using the slide-wheel. In the next step, we apply PCA to the data (spike vector) using the first 4 principal components for spike representation. According to the program's calculations, the total amount of variance explained by these 4 principal components reaches 80%. The program also informs us that the total number of spikes is 1196.

Under the first cluster, 211 spikes are classified.

The program identifies 244 spikes as noise. Under the noise class, all spike-overlapping activity is classified as non-fit to any proper cluster.

In order to exploit the above changes building the patched version of nev2lkit, we have issued a patch which is available for download here. (Please do note that nev2lkit requires Qt 3.* and Qwt 4.* to compile properly)

NEW! Precompiled binaries

o Mac OS X version 10.5 “Leopard” (Packed with Qt & Qwt libraries)
o Linux Ubuntu (x86) version 8.10 (Information about Qt & Qwt libraries here: linux_info.txt, linux_lib-pack.tar.gz)

The development of this work was made by Dimitrios A Adamos in the Laboratory of Animal Physiology, School of Biology, Faculty of Science, Aristotle University, Thessaloniki 54124, Greece.
This work was implemented in the context of PYTHAGORAS II (EPEAEK) project, a research program funded by the Greek Ministry of National Education and Religious Affairs.

1. Adamos D.A., Kosmidis E.K and Theophilidis G., "Performance evaluation of PCA-based spike sorting algorithms", Computer methods and programs in biomedicine (September 2008), vol. 91 (3) pp. 232-44.