Connectivity Measures for In Vitro Neuronal Cell Networks

Abstract

In this thesis, different connectivity measures are reviewed in detail in order to investigate what kind of information they provide, what are the advantages and limitations of them. Based on the literature review comparison, we selected three methods; Phase Lock Value (PLV), generalized Partial Directed Coherence (gPDC) and Transfer Entropy (TE). The selected methods were tested and evaluated with the data from human embryonic stem cell derived neuronal cell (hESC) networks which are cultured on MEAs. The analysis is divided into two parts: simulated connectivity signal studies and real MEA data analysis.The simulation study indicates that PLV method correctly recognized the connections, while gPDC provided unreliable results. TE provided the most detailed results only with few inaccuracies. Based on the simulation results, TE and PLV seem potential for further research on MEA signals. However, incoherent results were obtained in real MEA data analysis. For example, PLV claimed connections between signals measured from different wells. Based on the results, further research is needed in order to assess whether the incoherencies are influenced by the measurement environment, the methods themselves, or by the quality problem of signals in 6-well MEA.