Modeling Realistic Neuronal Activity in MEA Plates
Andalibi, Vafa (2015)
Master's Degree Programme in Biomedical Engineering
This publication is copyrighted. You may download, display and print it for Your own personal use. Commercial use is prohibited.
Julkaisun pysyvä osoite on
This thesis work is part of a project from Academy of Finland aiming at integrating biological components in sensor networks. The current integration goal considers neuronal cultures for achieving data processing. Due to the high capacity of neuronal cultures in parallel computation, the main assumptions of this project are that such integration will enable data processing that is not achievable with electrical components, and will reduce energy consumption. Within the scope of this project, the objective of this thesis is to develop realistic computational models of neuronal cultures plated on Multi-Electrode Arrays (MEAs). MEAs are integrated circuits used for stimulating cell cultures and recording their electrophysiological activity. Such models are used in the project for feasibility simulations and preliminary developments of bio-integrated systems (BIS). The contribution of this thesis is twofold: modeling plausible neural cultures on MEA, and analysis of the connectivity of neural networks. The first part contributes in gaining an in-depth understanding of the behavior of the neural network in MEA plate. A simulation framework is designed, implemented and used to simulate the neuronal activity in a MEA plate containing 1000 neurons. Using the implemented framework, it is now possible to simulate a MEA plate with many customizable parameters, e.g. MEA size, neuron size, type and morphology. The second part contributes with two implementations of a method for functional analysis of neural networks. Two GPU-accelerated algorithms of the Cox method were implemented with the CUDA platform. The Cox method is a proven robust method for the analysis of functional connectivity in networks. This method, formerly demanding a long time as well as consequent CPU power, can now run hundreds of times faster on CUDA-supported GPUs in personal computers.