Presentation

Search Abstracts | Symposia | Slide Sessions | Poster Sessions | Lightning Talks

EEG Hyper-Connectivity and high-gamma Differentiate PPA from Normal Aging

Poster C52 in Poster Session C, Wednesday, October 25, 10:15 am - 12:00 pm CEST, Espace Vieux-Port

Panteleimon Chriskos¹, Alexandros Afthinos², Jessica Gallegos², Christos Frantzidis^1,3, Brenda Rapp⁴, Panagiotis Bamidis¹, Kyrana Tsapkini^3,4; ¹Laboratory of Medical Physics and Digital Innovation, School of Medicine, Faculty of Health Sciences, Aristotle University of Thessaloniki, Thessaloniki, Greece, ²Department of Neurology, Johns Hopkins School of Medicine, Baltimore, MD, USA, ³School of Computer Science, University of Lincoln, Lincoln, United Kingdom, ⁴Department of Cognitive Science, Johns Hopkins University, Baltimore, MD, USA, ⁵Department of Physical Medicine and Rehabilitation, Johns Hopkins School of Medicine, Baltimore, MD, USA

Primary progressive aphasia (PPA) is a rare neurodegenerative syndrome characterized by progressive deterioration of language abilities. Diagnosis is assisted with medical imaging methods such as MRI or PET and lexical and/or language metrics. Significant effort has been made by many authors to automate the process of differentiating between healthy and PPA patients[1]. Low density EEG is of significant interest given its low cost and complexity, especially where equipment and specialized clinicians are scarce. However, very limited research is available regarding the use of EEG in PPA diagnosis. It is not clear which metrics characterize PPA and therefore differentiate PPA from normal aging. Here we will address this question using feature extraction in machine learning algorithms. We used resting state eyes-closed 8-channel (F7,F8,T7,T8,CP3,CP4,P5,P6) EEG recordings from 8 healthy elderly participants and 14 from PPA patients (9 lvPPA, 5 nfvPPA). The data was re-referenced using the common average re-referencing method, preprocessed to remove noise and the removal of (non-)linear trends, and consequently segmented into 8.192 second epochs. This resulted in a dataset with 361 healthy epochs and 294 PPA epochs. Functional connectivity was calculated using the Relative Wavelet Entropy (RWE) method, which is a method of constructing a synchronization matrix whose values represent the degree to which the energy distribution among the EEG rhythms (δ, θ, α, β and γ) are similar for each electrode pair. The relative energy ratios for each rhythm were also calculated for the whole EEG and for each electrode separately. Both the functional connectivity metrics and rhythm energy ratios were statistically analyzed using the Spearman rank correlation coefficient (ρ) to determine which features were statistically significant for distinguishing between the healthy control and patient groups. Functional hyper-connectivity in PPA compared to controls significantly differentiated the two groups in the following pairs: CP3-F7 (ρ=0.6195,p=1.0970e-07), CP3-T7 (ρ=-0.2375, p=7.4850e-10), F7-T7 (ρ=0.7083,p=7.3104e-10), F7-P5 (ρ=0.5951,p=5.3899e-06) and T7-P5 (ρ=0.1322, p=6.9477e-04). The EEG rhythm ratios that were statistically significant (p<0.001) were both slow rhythms as expected for a neurodegenerative disorder, such as the δ (whole EEG,F7,F8,T7,T8,CP3,P6), θ(T8,CP4,P5), α(F7,F8,P5,P6), and β(whole EEG, F7,T7,P5), as well as fast ones γ(F7, T7, T8, CP3, CP4, P5 and P6). The present study shows that EEG-based functional connectivity can reliably differentiate PPA from healthy controls and that EEG metrics can be used as a biomarker for PPA when other mediums are not available. Furthermore, we show for the first time, that PPA can be distinguished from healthy controls both in slow, as expected in neurodegeneration, but also in fast rhythms. EEG with its high temporal resolution offers further insight into the PPA syndrome. [1] Wilson, Stephen M., et al. "Connected speech production in three variants of primary progressive aphasia." Brain 133.7 (2010): 2069-2088.

Topic Areas: Disorders: Acquired, Computational Approaches