Analysis of changes in the brain neural electrical activity measured by the electroencephalogram (EEG) plays a crucial role in the area of brain disorder diagnostics. The elementary latent sources of the brain neural activity can be extracted by a tensor decomposition of continuously recorded multichannel EEG. Parallel factor analysis (PARAFAC) is a powerful approach for this purpose. However, the assumption of the same number of factors in each dimension of the PARAFAC model may be restrictive when applied to EEG data. In this article we discuss the potential benefits of an alternative tensor decomposition method – the Tucker model. We analyze situations, where in comparison to the PARAFAC solution, the Tucker model provides a more parsimonious representation of the EEG data decomposition. We show that this more parsimonious representation of EEG is achieved without reducing the ability to explain variance. We analyze EEG records of two patients after ischemic stroke and we focus on the extraction of specific sensorimotor oscillatory sources associated with motor imagery during neurorehabilitation training. Both models provided consistent results. The advantage of the Tucker model was a compact structure with only two spatial signatures reflecting the expected lateralized activation of the detected subject-specific sensorimotor rhythms.

The aim of this study is to implement an automatic system to detect the activation phases of Cyclic Alternating Pattern (CAP). CAP is a sleep phenomenon composed of consecutive sequences of activations and non-activations observed during the sleep time. Statistical Behavior of Local Extrema (SBLE) method is used for feature extraction and similarity index is used for decision making. SBLE is a symbolic technique that can characterize behavior of chaotic signals. The C4-A1 EEG lead of six healthy adult subjects was used. SBLE features are extracted from delta EEG band. By using this method and second by second comparison output of the system with sleep expert visual scoring, sensitivity of 75.83%, specificity of 81.30% and accuracy of 80.51% is achieved.

Today, a large number of people all around the world are affected or threatened by different kinds of sleep disorders, especially sleep apnea syndrome (SAS). Since EEG activities during sleep exhibit physiological and pathological situation of the brain, analyzing the oscillations during various sleep stages are of fundamental importance for the assessment of sleep disorders. On the other hand, recent developments in the fields of signal processing and pattern recognition have prompted sleep researchers to employ accurate and reliable computerized systems in diagnosis processes. The purpose of this paper was to develop a novel automated methodology for diagnosing apneic patients from healthy subjects using single-channel EEG signal. To this end, it was concentrated on the rapid eye movement (REM) sleep. A comprehensive time-domain based feature extraction scheme was proposed for extracting features from various EEG frequency bands of REM sleep. Uninformative features were eliminated by minimal redundancy maximal relevance (mRMR) feature selection algorithm. An ensemble learning method namely random undersampling boosting (RUSBoost) was then utilized for the purpose of classification. The effectiveness and generalizability of the proposed approach were evaluated using k-fold cross-validation technique. The experimental results demonstrate that the proposed scheme can be used as an efficient method for automated SAS screening with regard to promising classification performance in terms of accuracy, sensitivity, and specificity.

Functional dissociation of brain neural activity induced by opening or closing the eyes has been well established. However, how the temporal dynamics of the underlying neuronal modulations differ between these eye conditions during movement-related behaviours is less known. Using a robotic-assisted motor imagery brain-computer interface (MI BCI), we measured neural activity over the motor regions with electroencephalography (EEG) in a stroke survivor during his longitudinal rehabilitation training. We investigated lateralized oscillatory sensorimotor rhythm modulations while the patient imagined moving his hemiplegic hand with closed and open eyes to control an external robotic splint. In order to precisely identify the main profiles of neural activation affected by MI with eyes-open (MIEO) and eyes-closed (MIEC), a data-driven approach based on parallel factor analysis (PARAFAC) tensor decomposition was employed. Using the proposed framework, a set of narrow-band, subject-specific sensorimotor rhythms was identified; each of them had its own spatial and time signature. When MIEC trials were compared with MIEO trials, three key narrow-band rhythms whose peak frequencies centred at ∼8.0 Hz, ∼11.5 Hz, and ∼15.5 Hz, were identified with differently modulated oscillatory dynamics during movement preparation, initiation, and completion time frames. Furthermore, we observed that lower and higher sensorimotor oscillations represent different functional mechanisms within the MI paradigm, reinforcing the hypothesis that rhythmic activity in the human sensorimotor system is dissociated. Leveraging PARAFAC, this study achieves remarkable precision in estimating latent sensorimotor neural substrates, aiding the investigation of the specific functional mechanisms involved in the MI process.