Parkinson’s Disease Detection from Resting-State EEG Signals Using Common Spatial Pattern, Entropy, and Machine Learning Techniques

Abstract

Parkinson’s disease (PD) is a very common brain abnormality that affects people all over the world. Early detection of such abnormality is critical in clinical diagnosis in order to prevent disease progression. Electroencephalography (EEG) is one of the most important PD diagnostic tools since this disease is linked to the brain. In this study, novel efficient common spatial pattern-based approaches for detecting Parkinson’s disease in two cases, off–medication and on–medication, are proposed. First, the EEG signals are preprocessed to remove major artifacts before spatial filtering using a common spatial pattern. Several features are extracted from spatially filtered signals using different metrics, namely, variance, band power, energy, and several types of entropy. Machine learning techniques, namely, random forest, linear/quadratic discriminant analysis, support vector machine, and k-nearest neighbor, are investigated to classify the extracted features. The impacts of frequency bands, segment length, and reduction number on the results are also investigated in this work. The proposed methods are tested using two EEG datasets: the SanDiego dataset (31 participants, 93 min) and the UNM dataset (54 participants, 54 min). The results show that the proposed methods, particularly the combination of common spatial patterns and log energy entropy, provide competitive results when compared to methods in the literature. The achieved results in terms of classification accuracy, sensitivity, and specificity in the case of off-medication PD detection are around 99%. In the case of on-medication PD, the results range from 95% to 98%. The results also reveal that features extracted from the alpha and beta bands have the highest classification accuracy.

Detection of Parkinson’s disease from EEG signals using discrete wavelet transform, different entropy measures, and machine learning techniques

“…The present study used the discrete wavelet transform with threshold entropy, sure entropy, or proposed T-Shannon entropy, achieving accuracy scores of 99.72, 99.66, and 99.89%, respectively (see Table 5). These results are superior to the results of 29,34,35 , and close to the result reported in 30 .…”

“…Our proposed methods are compared with existing state-of-the-art techniques to assess their effectiveness for off-PD versus HC classification. From the previous studies summarized in Table 1, we focus on studies that used the same dataset 29,30,34,35 . In 29 , first, the Gabor transformation was used to convert the EEG signals to spectrograms.…”

“…After that, statistical measurements were used to extract five features from these subbands, obtaining an accuracy of 96.13% using the least square SVM. The authors of 35 proposed a combination of common spatial pattern and entropy, obtaining an accuracy of 99.41% using the KNN classifier. The present study used the discrete wavelet transform with threshold entropy, sure entropy, or proposed T-Shannon entropy, achieving accuracy scores of 99.72, 99.66, and 99.89%, respectively (see Table 5).…”

Generalizable electroencephalographic classification of Parkinson’s Disease using deep learning

“…Another technique that has been common in deep learning classification of PD-EEG data is to use K-fold cross validation to show the robust classification ability. However, Aljalal et al noted that the combination of epoching and cross validation leads to an information leakage between the train and test sets since both sets will include epochs from the same subjects [6]. To highlight the effect of this information leakage, we performed an investigation using the previous approaches, and found that classification metrics may be dramatically exaggerated (by 20% or more) to values that are not representative of the classifier’s true performance on the test set.…”

“…Many of these studies report very high accuracies above 98%. [5 - 12] However, the methods employed in these articles may suffer from the limitation of information-leakage between the training and testing data [6]. This limitation can lead to large overestimation of classification ability on unseen, real-world data.…”