Network influence based classification and comparison of neurological conditions

Research output: Contribution to conferencePaper

10 Downloads (Pure)

Abstract

Variations in the influence of brain regions are used to classify neurological conditions by identifying eigenvector-based communities in connectivity matrices, generated from resting state functional magnetic resonance imaging scans. These communities capture the network influence of each brain region, revealing that the subjects with Alzheimer’s disease (AD) have a significantly lower degree of variation in their most influential brain regions when compared with healthy control (HC) and amnestic mild cognitive impairment (aMCI) subjects. Classification of subjects based on their pattern of influential regions is demonstrated with neural networks identifying HC, aMCI and AD subjects. The difference between these conditions are investigated by altering brain region influence so that a neural network changes a subject’s classification. This conversion is performed on healthy subjects changing to aMCI or AD, and for aMCI subjects changing to AD. The results highlight potential compensatory mechanisms that increase functional activity in certain regions for those with aMCI, such as in the right parahippocampal gyrus and regions in the default mode network, but these same regions experience significant decline in those that convert from aMCI to AD.
Original languageEnglish
Number of pages13
Publication statusPublished - 10 Dec 2019
EventComplex Networks 2019: The 8th International Conference on Complex Networks and their Applications - Lisbon, Portugal
Duration: 10 Dec 201912 Dec 2019
https://www.complexnetworks.org/

Conference

ConferenceComplex Networks 2019
CountryPortugal
CityLisbon
Period10/12/1912/12/19
Internet address

Keywords

  • functional connectivity
  • community detection
  • eigenvector
  • dementia

Fingerprint Dive into the research topics of 'Network influence based classification and comparison of neurological conditions'. Together they form a unique fingerprint.

Cite this