MEArec: A Fast and Customizable Testbench Simulator for Ground-truth Extracellular Spiking Activity

Area of Science:

• Neuroscience
• Computational Neuroscience
• Signal Processing

Background:

• Spike sorting is essential for identifying single neuron activity from extracellular recordings.
• Validating spike sorting algorithms is challenging due to the unsupervised nature of the problem and lack of universal performance metrics.
• Existing ground-truth methods are limited by the number of simultaneously recorded neurons.

Purpose of the Study:

• To introduce MEArec, a Python-based software for simulating extracellular recordings.
• To provide a flexible and efficient tool for generating realistic neural data for spike sorting evaluation.
• To establish a common testbench for the development and benchmarking of spike sorting algorithms.

Main Methods:

• MEArec simulates extracellular signals on customizable electrode designs.
• The software can replicate challenging spike sorting scenarios like bursting, overlapping events, and signal drift.
• It generates simulated ground-truth data for quantitative evaluation of spike sorting performance.

Main Results:

• MEArec enables the creation of diverse and complex simulated extracellular recordings.
• The tool facilitates the generation of problematic aspects commonly encountered in real neural data.
• It offers a reproducible method for assessing spike sorting algorithm performance.

Conclusions:

• MEArec provides a valuable resource for the neuroscience community to rigorously evaluate spike sorting methods.
• The software promotes the development of more accurate and reliable spike sorting algorithms.
• This simulation approach offers a scalable and cost-effective alternative to experimental ground-truth recordings.