A New dispersion entropy and fuzzy logic system methodology for automated classification of dementia stages using electroencephalograms

Performance Evaluation of Deep, Shallow and Ensemble Machine Learning Methods for the Automated Classification of Alzheimer's Disease

Artificial intelligence (AI)-based approaches are crucial in computer-aided diagnosis (CAD) for various medical applications. Their ability to quickly and accurately learn from complex data is remarkable. Deep learning (DL) models have shown promising results in accurately classifying Alzheimer's disease (AD) and its related cognitive states, Early Mild Cognitive Impairment (EMCI) and Late Mild Cognitive Impairment (LMCI), along with the healthy conditions known as Cognitively Normal (CN). This offers valuable insights into disease progression and diagnosis. However, certain traditional machine learning (ML) classifiers perform equally well or even better than DL models, requiring less training data. This is particularly valuable in CAD in situations with limited labeled datasets. In this paper, we propose an ensemble classifier based on ML models for magnetic resonance imaging (MRI) data, which achieved an impressive accuracy of 96.52%. This represents a 3-5% improvement over the best individual classifier. We evaluated popular ML classifiers for AD classification under both data-scarce and data-rich conditions using the Alzheimer's Disease Neuroimaging Initiative and Open Access Series of Imaging Studies datasets. By comparing the results to state-of-the-art CNN-centric DL algorithms, we gain insights into the strengths and weaknesses of each approach. This work will help users to select the most suitable algorithm for AD classification based on data availability.

Alzheimer's Disease Evaluation Through Visual Explainability by Means of Convolutional Neural Networks

Background and Objective: Alzheimer's disease is nowadays the most common cause of dementia. It is a degenerative neurological pathology affecting the brain, progressively leading the patient to a state of total dependence, thus creating a very complex and difficult situation for the family that has to assist him/her. Early diagnosis is a primary objective and constitutes the hope of being able to intervene in the development phase of the disease. Methods: In this paper, a method to automatically detect the presence of Alzheimer's disease, by exploiting deep learning, is proposed. Five different convolutional neural networks are considered: ALEX_NET, VGG16, FAB_CONVNET, STANDARD_CNN and FCNN. The first two networks are state-of-the-art models, while the last three are designed by authors. We classify brain images into one of the following classes: non-demented, very mild demented and mild demented. Moreover, we highlight on the image the areas symptomatic of Alzheimer presence, thus providing a visual explanation behind the model diagnosis. Results: The experimental analysis, conducted on more than 6000 magnetic resonance images, demonstrated the effectiveness of the proposed neural networks in the comparison with the state-of-the-art models in Alzheimer's disease diagnosis and localization. The best results in terms of metrics are the best with STANDARD_CNN and FCNN with accuracy, precision and recall between 98% and 95%. Excellent results also from a qualitative point of view are obtained with the Grad-CAM for localization and visual explainability. Conclusions: The analysis of the heatmaps produced by the Grad-CAM algorithm shows that in almost all cases the heatmaps highlight regions such as ventricles and cerebral cortex. Future work will focus on the realization of a network capable of analyzing the three anatomical views simultaneously.

Self-Supervised Learning for Electroencephalography

Decades of research have shown machine learning superiority in discovering highly nonlinear patterns embedded in electroencephalography (EEG) records compared with conventional statistical techniques. However, even the most advanced machine learning techniques require relatively large, labeled EEG repositories. EEG data collection and labeling are costly. Moreover, combining available datasets to achieve a large data volume is usually infeasible due to inconsistent experimental paradigms across trials. Self-supervised learning (SSL) solves these challenges because it enables learning from EEG records across trials with variable experimental paradigms, even when the trials explore different phenomena. It aggregates multiple EEG repositories to increase accuracy, reduce bias, and mitigate overfitting in machine learning training. In addition, SSL could be employed in situations where there is limited labeled training data, and manual labeling is costly. This article: 1) provides a brief introduction to SSL; 2) describes some SSL techniques employed in recent studies, including EEG; 3) proposes current and potential SSL techniques for future investigations in EEG studies; 4) discusses the cons and pros of different SSL techniques; and 5) proposes holistic implementation tips and potential future directions for EEG SSL practices.

Computer aided progression detection model based on optimized deep LSTM ensemble model and the fusion of multivariate time series data

Alzheimer's disease (AD) is the most common form of dementia. Early and accurate detection of AD is crucial to plan for disease modifying therapies that could prevent or delay the conversion to sever stages of the disease. As a chronic disease, patient's multivariate time series data including neuroimaging, genetics, cognitive scores, and neuropsychological battery provides a complete profile about patient's status. This data has been used to build machine learning and deep learning (DL) models for the early detection of the disease. However, these models still have limited performance and are not stable enough to be trusted in real medical settings. Literature shows that DL models outperform classical machine learning models, but ensemble learning has proven to achieve better results than standalone models. This study proposes a novel deep stacking framework which combines multiple DL models to accurately predict AD at an early stage. The study uses long short-term memory (LSTM) models as base models over patient's multivariate time series data to learn the deep longitudinal features. Each base LSTM classifier has been optimized using the Bayesian optimizer using different feature sets. As a result, the final optimized ensembled model employed heterogeneous base models that are trained on heterogeneous data. The performance of the resulting ensemble model has been explored using a cohort of 685 patients from the University of Washington's National Alzheimer's Coordinating Center dataset. Compared to the classical machine learning models and base LSTM classifiers, the proposed ensemble model achieves the highest testing results (i.e., 82.02, 82.25, 82.02, and 82.12 for accuracy, precision, recall, and F1-score, respectively). The resulting model enhances the performance of the state-of-the-art literature, and it could be used to build an accurate clinical decision support tool that can assist domain experts for AD progression detection.

Hybrid Deep Learning Approach for Stress Detection Using Decomposed EEG Signals

Stress has an impact, not only on a person's physical health, but also on the ability to perform at the workplace in daily life. The well-established relation between psychological stress and its pathogeneses highlights the need for detecting psychological stress early, in order to prevent disease advancement and to save human lives. Electroencephalography (EEG) signal recording tools are widely used to collect these psychological signals/brain rhythms in the form of electric waves. The aim of the current research was to apply automatic feature extraction to decomposed multichannel EEG recordings, in order to efficiently detect psychological stress. The traditional deep learning techniques, namely the convolution neural network (CNN), long short-term memory (LSTM), bidirectional long short-term memory (BiLSTM), gated recurrent unit (GRU) and recurrent neural network (RNN) models, have been frequently used for stress detection. A hybrid combination of these techniques may provide improved performance, and can handle long-term dependencies in non-linear brain signals. Therefore, this study proposed an integration of deep learning models, called DWT-based CNN, BiLSTM, and two layers of a GRU network, to extract features and classify stress levels. Discrete wavelet transform (DWT) analysis was used to remove the non-linearity and non-stationarity from multi-channel (14 channel) EEG recordings, and to decompose them into different frequency bands. The decomposed signals were utilized for automatic feature extraction using the CNN, and the stress levels were classified using BiLSTM and two layers of GRU. This study compared five combinations of the CNN, LSTM, BiLSTM, GRU and RNN models with the proposed model. The proposed hybrid model performed better in classification accuracy compared to the other models. Therefore, hybrid combinations are appropriate for the clinical intervention and prevention of mental and physical problems.

Refined composite multiscale fluctuation-based dispersion Lempel-Ziv complexity for signal analysis

Dispersion Lempel-Ziv complexity (DLZC) and multiscale DLZC (MDLZC) are very recently introduced complexity indicators to quantify the dynamic change of time series in acoustics signals. They introduce the mapping steps of dispersion entropy (DE), which can effectively identify time series with different characteristics, but ignore the fluctuation information and have poor stability. In order to overcome these shortcomings, this paper firstly adds fluctuation information to DLZC and proposes fluctuation-based DLZC (FDLZC) as an alternative to the classical time series complexity index, followed by introducing an improved coarse-graining operation to propose the refined composite multiscale FDLZC (RCMFDLZC), which increases the number of features and also ensures the stability of FDLZC, and finally select the subsequence containing the most information by the minimum redundancy maximum relevance (mRMR) feature selection algorithm for subsequent experiments. The experimental results show that the extracted RCMFDLZC features have the strongest separability and better clustering effect in both bearing fault signals and ship radiated noise signals, and the RCMFDLZC-based signal analysis method also has higher recognition rate compared with other methods in bearing fault diagnosis and ship signal classification.

De-Noising of Magnetotelluric Signals by Discrete Wavelet Transform and SVD Decomposition

Magnetotelluric (MT) sounding data can easily be damaged by various types of noise, especially in industrial areas, where the quality of measured data is poor. Most traditional de-noising methods are ineffective to the low signal-to-noise ratio of data. To solve the above problem, we propose the use of a de-noising method for the detection of noise in MT data based on discrete wavelet transform and singular value decomposition (SVD), with multiscale dispersion entropy and phase space reconstruction carried out for pretreatment. No “over processing” takes place in the proposed method. Compared with wavelet transform and SVD decomposition in synthetic tests, the proposed method removes the profile of noise more completely, including large-scale noise and impulse noise. For high levels or low levels of noise, the proposed method can increase the signal-to-noise ratio of data more obviously. Moreover, application to the field MT data can prove the performance of the proposed method. The proposed method is a feasible method for the elimination of various noise types and can improve MT data with high noise levels, obtaining a recovery in the response. It can improve abrupt points and distortion in MT response curves more effectively than the robust method can.

Control of Transcranial Direct Current Stimulation Duration by Assessing Functional Connectivity of Near-Infrared Spectroscopy Signals

Transcranial direct current stimulation (tDCS) has been shown to create neuroplasticity in healthy and diseased populations. The control of stimulation duration by providing real-time brain state feedback using neuroimaging is a topic of great interest. This study presents the feasibility of a closed-loop modulation for the targeted functional network in the prefrontal cortex. We hypothesize that we cannot improve the brain state further after reaching a specific state during a stimulation therapy session. A high-definition tDCS of 1[Formula: see text]mA arranged in a ring configuration was applied at the targeted right prefrontal cortex of 15 healthy male subjects for 10[Formula: see text]min. Functional near-infrared spectroscopy was used to monitor hemoglobin chromophores during the stimulation period continuously. The correlation matrices obtained from filtered oxyhemoglobin were binarized to form subnetworks of short- and long-range connections. The connectivity in all subnetworks was analyzed individually using a new quantification measure of connectivity percentage based on the correlation matrix. The short-range network in the stimulated hemisphere showed increased connectivity in the initial stimulation phase. However, the increase in connection density reduced significantly after 6[Formula: see text]min of stimulation. The short-range network of the left hemisphere and the long-range network gradually increased throughout the stimulation period. The connectivity percentage measure showed a similar response with network theory parameters. The connectivity percentage and network theory metrics represent the brain state during the stimulation therapy. The results from the network theory metrics, including degree centrality, efficiency, and connection density, support our hypothesis and provide a guideline for feedback on the brain state. The proposed neuro-feedback scheme is feasible to control the stimulation duration to avoid overdosage.

Assessment of Dispersion and Bubble Entropy Measures for Enhancing Preterm Birth Prediction Based on Electrohysterographic Signals

One of the remaining challenges for the scientific-technical community is predicting preterm births, for which electrohysterography (EHG) has emerged as a highly sensitive prediction technique. Sample and fuzzy entropy have been used to characterize EHG signals, although they require optimizing many internal parameters. Both bubble entropy, which only requires one internal parameter, and dispersion entropy, which can detect any changes in frequency and amplitude, have been proposed to characterize biomedical signals. In this work, we attempted to determine the clinical value of these entropy measures for predicting preterm birth by analyzing their discriminatory capacity as an individual feature and their complementarity to other EHG characteristics by developing six prediction models using obstetrical data, linear and non-linear EHG features, and linear discriminant analysis using a genetic algorithm to select the features. Both dispersion and bubble entropy better discriminated between the preterm and term groups than sample, spectral, and fuzzy entropy. Entropy metrics provided complementary information to linear features, and indeed, the improvement in model performance by including other non-linear features was negligible. The best model performance obtained an F1-score of 90.1 ± 2% for testing the dataset. This model can easily be adapted to real-time applications, thereby contributing to the transferability of the EHG technique to clinical practice.