Development of analysis pipelines in MATLAB for electrophysiological signals in vitro and in vivo

This thesis is focused on my internship in Prof. Stefano Vassanelli “NeuroChip” Lab at the University of Padova. I used MATLAB to perform the data analysis of electrophysiological signals collected by the team of Prof. Stefano Vassanelli for two different EU projects. The goal of the internship was to apply electrophysiological analysis methods firsthand, both in vitro and in vivo. In the first part of my internship, I analyzed signals recorded from neuronal cultures in vitro. In particular, these data were obtained from cultures grown on two different types of Micro-Electrode Arrays (MEAs) equipped with vertical nanowires (NWs). The microchips were being tested for the EU project NEUREKA. The microchips of the first type are called Nanowire Electrode Array (NEA) and they allow both the recording and the stimulation of neurons from the same chip. The analysis involves filtering and temporal alignment of the signals recorded by patch-clamp technique with those obtained directly from the microchip. The microchips of the second type, on the other hand, do not allow stimulation and recording at the same time and are called “Bulk”. In this case, data analysis is carried out only on signals recorded by patch-clamp. Each response is classified according to the characteristics of the stimulation, which occurs through the microchip itself, and the presence of action potentials and/or subthreshold potentials. The second part of the internship was focused on the analysis of electrophysiological signals, in particular Local Field Potentials (LFPs), recorded from the mouse somatosensory cortex in vivo, using a multichannel probe (Atlas Neuro). The data were collected for the MSCA EU project “GRACE”, carried out by Dr. Claudia Cecchetto in UNIPD and OIST (Okinawa, Japan). For the analysis of LFPs acquired during epochs with no stimulations (i.e., spontaneous activity), a preliminary frequency analysis is performed. Then, time-frequency analysis follows, including theorical references for both the short-time Fourier transform and the wavelet transform. Cross-correlation of the channels is also performed, in order to hypothesize a distinction between channels based on the layer/barrel they belong to and study the temporal relationship between signals recorded at different depths. A distinction is also made considering specific frequency bands separately. Regarding the analysis of neuronal responses evoked by whisker stimulation using air-puff, some characteristic parameters are extracted from the evoked responses (e.g., the amplitude and duration of negative and positive peaks). Then, time-frequency analysis is included in order to investigate whether the spectrogram and the scalogram show significant changes when the stimulation happens. Lastly, cross-correlation is performed to study the directionality of the evoked response in the cortex, analyzing the temporal relationship between the channels. By means of the maximum cross-correlation matrices, obtained from both spontaneous and stimulated activity, a probable subdivision of the channels into at least two barrels and three or four different layers is recognized. Additionally, it appears that the temporal relationship between superficial and deeper channels changes depending on the frequency band that is considered.