Early diagnosis of Parkinson’s disease using EEG, machine learning and partial directed coherence

Oliveira, Ana Paula S.; Santana, Maíra Araújo de; Andrade, Maria Karoline S.; Gomes, Juliana Carneiro; Rodrigues, Michele; Santos, Wellington Pinheiro dos

doi:10.1007/s42600-020-00072-w

Cited by 36 publications

(14 citation statements)

References 88 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Quantitative analyses (both fast Fourier transformation (FFT) and coherence analysis) were performed using the Brainstorm software [ 42 ] on MATLAB compiler runtime R2015b (MCR v9.0). The amplitude for each frequency band was calculated through the FFT and the functional connectivity through the coherence method ( Retrieved on 13 April 2021) previously used [ 43 ]. Relative amplitude was calculated as a percentage for each frequency band from the total absolute amplitude spectrum (from 1 to 32 Hz).…”

Section: Methodsmentioning

confidence: 99%

Fibromyalgia Detection Based on EEG Connectivity Patterns

Martín-Brufau

Gómez

Sánchez-Sánchez-Rojas

et al. 2021

JCM

View full text Add to dashboard Cite

Objective: The identification of a complementary test to confirm the diagnosis of FM. The diagnosis of fibromyalgia (FM) is based on clinical features, but there is still no consensus, so patients and clinicians might benefit from such a test. Recent findings showed that pain lies in neuronal bases (pain matrices) and, in the long term, chronic pain modifies the activity and dynamics of brain structures. Our hypothesis is that patients with FM present lower levels of brain activity and therefore less connectivity than controls. Methods: We registered the resting state EEG of 23 patients with FM and compared them with 23 control subjects’ resting state recordings from the PhysioBank database. We measured frequency, amplitude, and functional connectivity, and conducted source localization (sLORETA). ROC analysis was performed on the resulting data. Results: We found significant differences in brain bioelectrical activity at rest in all analyzed bands between patients and controls, except for Delta. Subsequent source analysis provided connectivity values that depicted a distinct profile, with high discriminative capacity (between 91.3–100%) between the two groups. Conclusions: Patients with FM show a distinct neurophysiological pattern that fits with the clinical features of the disease.

show abstract

Section: Methodsmentioning

confidence: 99%

Fibromyalgia Detection Based on EEG Connectivity Patterns

Martín-Brufau

Gómez

Sánchez-Sánchez-Rojas

et al. 2021

JCM

View full text Add to dashboard Cite

show abstract

“…These studies are summarized in Table 1. Eight out of ten automated PD detection studies in Table 1 proposed conventional machine-learning models [21,27,[29][30][31][32][33][34], and half of these studies employed a supportvector-machine (SVM) classifier. The highest classification accuracy obtained using a machine-learning methodology was by Yuvaraj et al [34].…”

Section: Related Workmentioning

confidence: 99%

GaborPDNet: Gabor Transformation and Deep Neural Network for Parkinson’s Disease Detection Using EEG Signals

et al. 2021

View full text Add to dashboard Cite

Parkinson’s disease (PD) is globally the most common neurodegenerative movement disorder. It is characterized by a loss of dopaminergic neurons in the substantia nigra of the brain. However, current methods to diagnose PD on the basis of clinical features of Parkinsonism may lead to misdiagnoses. Hence, noninvasive methods such as electroencephalographic (EEG) recordings of PD patients can be an alternative biomarker. In this study, a deep-learning model is proposed for automated PD diagnosis. EEG recordings of 16 healthy controls and 15 PD patients were used for analysis. Using Gabor transform, EEG recordings were converted into spectrograms, which were used to train the proposed two-dimensional convolutional neural network (2D-CNN) model. As a result, the proposed model achieved high classification accuracy of 99.46% (±0.73) for 3-class classification (healthy controls, and PD patients with and without medication) using tenfold cross-validation. This indicates the potential of proposed model to simultaneously automatically detect PD patients and their medication status. The proposed model is ready to be validated with a larger database before implementation as a computer-aided diagnostic (CAD) tool for clinical-decision support.

show abstract

“…In recent years, machine learning (ML) and deep learning algorithms have been used to identify EEG-based neuromarkers in PD, including early-stage markers [30][31][32][33][34][35]. These efforts relied mainly on resting-state EEG.…”

Section: Introductionmentioning

confidence: 99%

Identification of an early-stage Parkinson’s disease neuromarker using event-related potentials, brain network analytics and machine-learning

et al. 2022

View full text Add to dashboard Cite

Objective The purpose of this study is to explore the possibility of developing a biomarker that can discriminate early-stage Parkinson’s disease from healthy brain function using electroencephalography (EEG) event-related potentials (ERPs) in combination with Brain Network Analytics (BNA) technology and machine learning (ML) algorithms. Background Currently, diagnosis of PD depends mainly on motor signs and symptoms. However, there is need for biomarkers that detect PD at an earlier stage to allow intervention and monitoring of potential disease-modifying therapies. Cognitive impairment may appear before motor symptoms, and it tends to worsen with disease progression. While ERPs obtained during cognitive tasks performance represent processing stages of cognitive brain functions, they have not yet been established as sensitive or specific markers for early-stage PD. Methods Nineteen PD patients (disease duration of ≤2 years) and 30 healthy controls (HC) underwent EEG recording while performing visual Go/No-Go and auditory Oddball cognitive tasks. ERPs were analyzed by the BNA technology, and a ML algorithm identified a combination of features that distinguish early PD from HC. We used a logistic regression classifier with a 10-fold cross-validation. Results The ML algorithm identified a neuromarker comprising 15 BNA features that discriminated early PD patients from HC. The area-under-the-curve of the receiver-operating characteristic curve was 0.79. Sensitivity and specificity were 0.74 and 0.73, respectively. The five most important features could be classified into three cognitive functions: early sensory processing (P50 amplitude, N100 latency), filtering of information (P200 amplitude and topographic similarity), and response-locked activity (P-200 topographic similarity preceding the motor response in the visual Go/No-Go task). Conclusions This pilot study found that BNA can identify patients with early PD using an advanced analysis of ERPs. These results need to be validated in a larger PD patient sample and assessed for people with premotor phase of PD.

show abstract

Early diagnosis of Parkinson’s disease using EEG, machine learning and partial directed coherence

Cited by 36 publications

References 88 publications

Fibromyalgia Detection Based on EEG Connectivity Patterns

Fibromyalgia Detection Based on EEG Connectivity Patterns

GaborPDNet: Gabor Transformation and Deep Neural Network for Parkinson’s Disease Detection Using EEG Signals

Identification of an early-stage Parkinson’s disease neuromarker using event-related potentials, brain network analytics and machine-learning

Contact Info

Product

Resources

About