IDCOM seminar series

Location: 

Classroom 4, Hudson Beare Building

Date: 

Monday, May 11, 2015 - 13:00 to 14:00

Keith Smith

University of Edinburgh - IDCOM & AlzScotDRC

New developments in Network Theory for the Preclinical Detection of Alzheimer's Disease and Beyond

Abstract:

Alzheimer's Disease (AD) is the most common form of dementia in the world. By the time it is currently diagnosable, significant irreversible brain damage has already occurred. Thus new methods for detection at preclinical stages which are applicable in large populations are required. In order to tackle this problem our research proposes network analysis of Electroencephalogram (EEG) brain signals from performance of Visual Short-Term Memory (VSTM) tasks as a methodology for an early biomarker of AD.
Network theory is now a standard mathematical framework for understanding complex, interdependent phenomena. A network is an abstract representation consisting of a set of nodes with connections formed between them. In application to brain imaging and signal processing we obtain weighted complete networks, where weighted connections exist between every pair of nodes. It is desirable to binarise these networks using a threshold in order to nullify the many spurious low weights and simplify analysis. However, selection of a threshold has so far been arbitrary. This leads to study-by-study differences, confounding any comparison attempts. Here we report on a new way of approaching conceptual groundings of network theory. This leads to a solution to the problem of arbitrary threshold selection and brings fresh insights for weighted complete networks. We present our findings of this new threshold, the Cluster-Span Threshold, applied to EEG data of young healthy adults performing different VSTM tasks, showing promising results in our research into detection of AD.

Info: Keith Smith is in his first year of an interdisciplinary PhD in Engineering (IDCOM & AlzScotDRC). His research is focused on developing a Network Theory framework for the early detection of Alzheimer?s Disease from EEG functional connectivity. Future research plans includes work on a new criteria for understanding different types of networks and developing techniques for analysing temporal dynamics of EEG functional brain networks utilising concepts from Signal Processing on Graphs and Multi-layer Networks.

Event Contact Name: 

IDCoM seminar team

Event Contact Email: