Resting state EEG as a biomarker of Parkinson’s disease: Influence of measurement conditions

Abstract

Resting state electroencephalographic (EEG) recording could provide cost-effective means to aid in the detection of neurological disorders such as Parkinson's disease (PD). We examined how many electrodes are needed for classification of PD based on EEG, which electrode locations provide most value for classification, and whether data recorded eyes open or closed yield comparable results. We used a nested cross-validated classifier which included a budgetbased search algorithm for selecting the optimal electrodes for classification. By iterating over variable budgets, we show that with eyes open recording, only 10 electrodes, localized over motor and occipital areas enable relatively accurate classification (AUC = .82) between PD patients (N=20) and age-matched healthy control participants (N=20). Classification accuracy only slightly increased when all 64 electrodes were included (AUC = .85). With the data recorded eyes closed, classification was not statistically significantly above chance even with full set of 64 electrodes (AUC = .55). These results show that classification based on small number of EEG electrodes is a promising tool for classifying PD, but measurement conditions and electrode locations can have a significant effect on classifier performance.

The Effects of Non-Invasive Brain Stimulation on Quantitative EEG in Patients With Parkinson's Disease: A Systematic Scoping Review

“…Concerning the eyes condition, previous studies were able to differentiate and classify patients with PD and healthy controls at rest with the eyes closed and during tasks with eyes opened (47)(48)(49)(50). On the other hand, Railo et al (51) demonstrated that patients with PD in the initial to intermediate state can be classified with relatively high sensitivity using EEG data recorded at rest with eyes open with about 10 electrodes, located over the motor and occipital areas. Contrary, the classification was not possible with the eyes closed (51).…”

“…On the other hand, Railo et al ( 51 ) demonstrated that patients with PD in the initial to intermediate state can be classified with relatively high sensitivity using EEG data recorded at rest with eyes open with about 10 electrodes, located over the motor and occipital areas. Contrary, the classification was not possible with the eyes closed ( 51 ). At present, it should be recommended to record EEG both with eyes opened and closed, whenever possible to test if the NIBS-induced changes are detectable at one condition or another or in both conditions.…”

Evaluating Effects of Resting-State Electroencephalography Data Pre-Processing on a Machine Learning Task for Parkinson’s Disease

“…One collected by direct measurement at The Canberra Hospital (TCH) in Australia, and two were publicly available data sets provided by the University of New Mexico (UNM), USA, and the University of Turku (UTU), Finland. [1] Table 1. Details of the three data sets used in this study.…”

“…One collected by direct measurement at The Canberra Hospital (TCH) in Australia, and two were publicly available data sets provided by the University of New Mexico (UNM), USA, and the University of Turku (UTU), Finland. [1] Data were pre-processed using MATLAB R2018b and the MATLAB toolbox EEGLAB 2020.0,[6] following the recommendations of the Swartz Centre of Computational Neuroscience and Makoto Miyakoshi. [7] Data were re-referenced to the average of the electrodes and down-sampled to 128 Hz.…”

“…Resting-state electroencephalography (RSEEG) is a method under consideration as a potential biomarker that could support early and accurate diagnosis of Parkinson's disease (PD), [1,2] as it can differentiate people with PD (PwPD) from controls. [3] In an effort to create diagnostic models, researchers have applied machine learning (ML) methods: [4] in most cases, RSEEG data are represented as features (e.g.…”