University of Twente Student Theses

Login

An in silico description of an alzheimer network

Rotgerink, J.L. (2013) An in silico description of an alzheimer network.

[img] PDF
2MB
Abstract:Alzheimer’s disease (AD) is the most well known form of dementia. Already in 1907 Alois Alzheimer published a first paper about the symptoms of this disease, of which we know that it afflicts one in three people above 85. By now it is broadly accepted that indeed the loss of short-term memory is among the first symptoms of AD, followed by behavioural changes and difficulties with motor activities. In the end, Alzheimer is a terminal disease, and with its huge direct impact on caregivers and financial impact on society it is an essential topic in several research fields. Additionally, by the increasing average age of current society it only becomes more urgent. Among these fields of research, computational neuroscience can be helpful as a first step in deciphering phenomena that occur with the appearance of Alzheimer’s disease. For many years after the first descriptions about AD there was no clue about a possible cause. Nowadays it is known that the appearance of the disease is mediated by a protein called Amyloid beta (Aβ). The exact role of Aβ remains unknown, though a prominent hypothesis is that it affects several presynaptic processes, thereby suppressing neuron efficiency. We model networks containing up to 600 neurons as computational networks in which the vertices represent a neuron. The internal dynamics of the neuron are governed by the so-called Izhikevich model. The connections represent the synapses between neurons. In this way, the behaviour of neural networks can be simulated by using computational mathematics. In this thesis a computational framework designed to map experimental data to synaptic parameters is constructed. In this way, having in vitro measurements of Aβ-treated neuronal networks and untreated control networks, the existence of significant differences in the synaptic properties of the two types of networks can be systematically investigated. Taking our cues from previous research by Philip Hahn & Cameron McIntyre [2010], we attempt to use a Gauss-Newton least mean square error optimisation. The computational framework then extracts the first two statistical moments, Mean Firing Rate (MFR) and Coefficient of Variation (CV), from measurements. This Gauss-Newton method subsequently searches for synaptic properties yielding networks with comparable MFR and CV. We conclude that this Gauss-Newton method should be handled with great care. In particular, we find that, for our particular model the algorithm either diverges or converges at a poor rate. Besides this statement and giving an overview on how varying the synaptic parameters affects the behaviour of neural networks, this thesis gives elaborate suggestions for future computational research on Alzheimer’s disease.
Item Type:Essay (Master)
Faculty:EEMCS: Electrical Engineering, Mathematics and Computer Science
Subject:33 physics
Programme:Applied Mathematics MSc (60348)
Link to this item:https://purl.utwente.nl/essays/63845
Export this item as:BibTeX
EndNote
HTML Citation
Reference Manager

 

Repository Staff Only: item control page