| dc.description.abstract | This thesis is focused on three objectives. The first objective was to develop a method that allows automatic assessment of stem cell development conditions with the use of impedance spectroscopy and machine learning, the second objective was the study of dielectric properties of neurotransmitter GABA (γ-aminobutyric acid) by using dielectric relaxation spectroscopy (DRS), and the final objective was to find a new method for automatic determination of the physiological GABA levels in aqueous solution using DRS and machine learning techniques. With the aim of finding an alternative treatment for neurodegenerative diseases, the European project Training4CRM proposed the development of an implantable device which uses the integrated human stem cells on chip to restore the lost brain functions for patients with neurodegenerative diseases, such as Parkinson's and epilepsy. Some of the neurodegenerative diseases are caused by lack of neurotransmitters in the brain. In this thesis, we focused specifically on studying the inhibitory neurotransmitter GABA. Impedance measurements on stem cells combined with machine learning methods were used to automatically determine the proliferation and differentiation processes that the stem cells undergo during their development. Thereafter, DRS was employed in the study of the conformation and dielectric properties of the GABA molecule under different laboratory conditions. Finally, a new approach was developed in detecting the physiological concentrations of GABA by using the DRS together with machine learning techniques. The results show that the machine learning method appears to be a promising tool for distinguishing cell development processes, proliferation and differentiation, with an accuracy of distinguishment results of 100%, 100% and 91% for three investigated stem cell lines respectively. Moreover, the environmental dependency of GABA dielectric properties investigated in this thesis has given us new insights about the GABA molecule. More specifically about how the GABA molecules interact with each other as well as with other surrounding molecules. In addition, the measurements of GABA dissolved in cell medium showed that the presence of other chemical substances has an impact on both the conformation of the GABA molecules and how the surrounding molecules influence the dielectric properties of the GABA molecule. Lastly, with the combination of DRS and machine learning algorithms, we successfully distinguished 6 different physiological GABA concentrations from each other with an accuracy of 99.2% by using one neural network model. | eng |