Analysis and spike detection of neural data from a CMOS MEA system

Abstract

This thesis is intended to deal with the problem of analysis and spike detection in neural data acquired from a CMOS Microelectrode Array (MEA) system. In order for this to be carried out a Graphical User Interface (GUI) application has been created. The data to analyze comes from an in-vitro recording and stimulation system which uses a 65 x 65 CMOS MEA and it is stored in HDF5 files with the .cmcr extension. The GUI application developed in Python is able to read these files and perform a number of tasks which facilitate the detection and visualization of neural activity within the tissue subjected to analysis. This is accomplished by interpreting the voltage measurements at every coordinate of the array over time for the search of Action Potentials (APs) or spikes. Once the spikes are detected the information is stored to be presented in different ways and a comparison is carried out to detect when and where most activity has taken place. The GUI application created enables the visualization of both raw and filtered data at a particular coordinate over time and showing the spikes that have been detected. Moreover, a second type of plot makes it possible to view the whole array filtered data for an exact time sample, as well as which coordinates present most neural activity. Finally, a number of tables show useful information about the pixel coordinates, time and height of these spikes.