Sahbi Chaibi

Sleep spindle and K-complex detection using tunable Q-factor wavelet transform and morphological component analysis.

A novel framework for joint detection of sleep spindles and K-complex events, two hallmarks of sleep stage S2, is proposed. Sleep electroencephalography (EEG) signals are split into oscillatory (spindles) and transient (K-complex) components. This decomposition is conveniently achieved by applying morphological component analysis (MCA) to a sparse representation of EEG segments obtained by the recently introduced discrete tunable Q-factor wavelet transform (TQWT). Tuning the Q-factor provides a convenient and elegant tool to naturally decompose the signal into an oscillatory and a transient component. The actual detection step relies on thresholding (i) the transient component to reveal K-complexes and (ii) the time-frequency representation of the oscillatory component to identify sleep spindles. Optimal thresholds are derived from ROC-like curves (sensitivity vs. FDR) on training sets and the performance of the method is assessed on test data sets. We assessed the performance of our method using full-night sleep EEG data we collected from 14 participants. In comparison to visual scoring (Expert 1), the proposed method detected spindles with a sensitivity of 83.18% and false discovery rate (FDR) of 39%, while K-complexes were detected with a sensitivity of 81.57% and an FDR of 29.54%. Similar performances were obtained when using a second expert as benchmark. In addition, when the TQWT and MCA steps were excluded from the pipeline the detection sensitivities dropped down to 70% for spindles and to 76.97% for K-complexes, while the FDR rose up to 43.62 and 49.09%, respectively. Finally, we also evaluated the performance of the proposed method on a set of publicly available sleep EEG recordings. Overall, the results we obtained suggest that the TQWT-MCA method may be a valuable alternative to existing spindle and K-complex detection methods. Paths for improvements and further validations with large-scale standard open-access benchmarking data sets are discussed.

A reliable approach to distinguish between transient with and without HFOs using TQWT and MCA.

Recent studies have reported that discrete high frequency oscillations (HFOs) in the range of 80-500Hz may serve as promising biomarkers of the seizure focus in humans. Visual scoring of HFOs is tiring, time consuming, highly subjective and requires a great deal of mental concentration. Due to the recent explosion of HFOs research, development of a robust automated detector is expected to play a vital role in studying HFOs and their relationship to epileptogenesis. Therefore, a handful of automated detectors have been introduced in the literature over the past few years. In fact, all the proposed methods have been associated with high false-positive rates, which essentially arising from filtered sharp transients like spikes, sharp waves and artifacts. In order to specifically minimize false positive rates and improve the specificity of HFOs detection, we proposed a new approach, which is a combination of tunable Q-factor wavelet transform (TQWT), morphological component analysis (MCA) and complex Morlet wavelet (CMW). The main findings of this study can be summarized as follows: The proposed method results in a sensitivity of 96.77%, a specificity of 85.00% and a false discovery rate (FDR) of 07.41%. Compared to this, the classical CMW method applied directly on the signals without pre-processing by TQWT-MCA achieves a sensitivity of 98.71%, a specificity of 18.75%, and an FDR of 29.95%. The proposed method may be considered highly accurate to distinguish between transients with and without HFOs. Consequently, it is remarkably reliable and robust for the detection of HFOs.

A Robustness Comparison of Two Algorithms Used for EEG Spike Detection.

Spikes and sharp waves recorded on scalp EEG may play an important role in identifying the epileptogenic network as well as in understanding the central nervous system. Therefore, several automatic and semi-automatic methods have been implemented to detect these two neural transients. A consistent gold standard associated with a high degree of agreement among neuroscientists is required to measure relevant performance of different methods. In fact, scalp EEG data can often be corrupted by a set of artifacts and are not always served as data of gold standard. For this reason, the use of intracerebral EEG data mixed with gaussian noise seems to best resemble the output of scalp EEG brain and serves as a consistent gold standard. In the present framework, we test the robustness of two important methods that have been previously used for the automatic detection of epileptiform transients (spikes and sharp waves). These methods are based respectively on Discrete Wavelet Transform (DWT) and Continuous Wavelet Transform (CWT). Our purpose is to elaborate a comparative study in terms of sensitivity and selectivity changes via the decrease of Signal to Noise Ratio (SNR), which is ranged from 10 dB up to -10 dB. The results demonstrate that, DWT approach turns to be more stable in terms of sensitivity, and it successfully follows the detection of relevant spikes with the decrease of SNR. However, CWT-based approach remains more stable in terms of selectivity, so that, it performs well in terms of rejecting false spikes compared to DWT approach.

Meet Spinky: An Open-Source Spindle and K-Complex Detection Toolbox Validated on the Open-Access Montreal Archive of Sleep Studies (MASS)

Sleep spindles and K-complexes are among the most prominent micro-events observed in electroencephalographic (EEG) recordings during sleep. These EEG microstructures are thought to be hallmarks of sleep-related cognitive processes. Although tedious and time-consuming, their identification and quantification is important for sleep studies in both healthy subjects and patients with sleep disorders. Therefore, procedures for automatic detection of spindles and K-complexes could provide valuable assistance to researchers and clinicians in the field. Recently, we proposed a framework for joint spindle and K-complex detection (Lajnef et al., 2015a) based on a Tunable Q-factor Wavelet Transform (TQWT; Selesnick, 2011a) and morphological component analysis (MCA). Using a wide range of performance metrics, the present article provides critical validation and benchmarking of the proposed approach by applying it to open-access EEG data from the Montreal Archive of Sleep Studies (MASS; O’Reilly et al., 2014). Importantly, the obtained scores were compared to alternative methods that were previously tested on the same database. With respect to spindle detection, our method achieved higher performance than most of the alternative methods. This was corroborated with statistic tests that took into account both sensitivity and precision (i.e., Matthew’s coefficient of correlation (MCC), F1, Cohen κ). Our proposed method has been made available to the community via an open-source tool named Spinky (for spindle and K-complex detection). Thanks to a GUI implementation and access to Matlab and Python resources, Spinky is expected to contribute to an open-science approach that will enhance replicability and reliable comparisons of classifier performances for the detection of sleep EEG microstructure in both healthy and patient populations.

Detection of High Frequency Oscillations (HFOs) in the 80–500 Hz range in epilepsy recordings using decision tree analysis

Discrete High Frequency Oscillations (HFOs) in the range of 80-500 Hz have recently received attention as a promising reliable biomarkers for epileptic activity, both in scalp EEG as well as in intracranial recordings. HFOs are often characterized by variable durations (10-100 ms) and rates of occurrence (17.5 ± 9.5 / min). The total duration of HFOs is extremely small compared to the entire length of the EEG signals to be analyzed which, in the case of intracerebral recordings, are generally acquired over several days and sometimes up to weeks. As a result, visual marking of HFOs events associated with large amounts of EEG data is extremely tedious, inevitably subjective and requires a great deal of mental concentration. Therefore, automatic detection of HFOs can be very useful to propel the clinical use of HFOs as biomarkers of epileptogenic tissue and is crucial when conducting large-scale investigations of HFO activity. As a first step towards robust and reliable automatic detection, we propose in this paper a new method for HFOs detection based on Decision Tree analysis. The performance and added value of the proposed method are evaluated by comparing it with five other previously proposed methods. The HFO detection performances were tested in terms of sensitivity, False Discovery Rate (FDR) and Area Under the ROC Curve (AUC). Our results demonstrate that the decision-tree approach yields low false detection (FDR=8.62 %) but that, in its current implementation, it is not highly sensitive to HFO events (sensitivity=66.96 %). Nevertheless some advantages of the method are discussed and paths for further improvements are outlined.

Classification of epileptic cerebral activity using robust features and support vector machines

Epileptic seizure detection requires the study of electroencephalogram (EEG) data. Visual marking of seizure onset in such EEG recordings is quite tedious, naturally subjective, extremely time consuming, and it may lead to inaccurate detection. Thus, the development of a robust framework for automatic seizure classification is necessary and can be very useful in epilepsy investigation. In this paper, a classical method has been improved. Our contribution includes the use of linear and non linear features which have been incorporated into the Support Vector Machines (SVM) classifier. Accordingly, the detection performance has been compared using both radial basis functions (RBF) and linear SVM kernels. Our main finding reveals that the system can correctly classify the EEG data with an average sensitivity of 99.68%, an average specificity of 99.81% and an average accuracy of 99.75%, while 100% of sensitivity, specificity and accuracy are also achieved in single-trial classification. A final comparison between the performance levels obtained with our method and those obtained with previous techniques is undertaken to prove the effectiveness of our method for seizure detection.

Separation of transient and oscillatory cereberal activities using over-complete rational dilation wavelt transforms

Many physiological signals such as the electroencephalogram EEG are composed of the superposition of oscillatory activities and transient activities. The oscillatory activities come from rhythmic patterns like delta 0-3 Hz, theta 4-7 Hz, alpha 8-12 Hz, beta 12-30 Hz, gamma 26-100 Hz, and HFOs: High frequency oscillations 80-500Hz. Whereas, the transient activities come from non-rhythmic brain activities like spikes, sharp waves, artifacts, and vertex waves of varying amplitude; shape; and duration. The problem is that the transient activities with different morphologies could overlap in both time and frequency domain with oscillatory patterns, that make the detection a difficult task at present. Visual identification of HFOs which represent an important biomarker of the seizure focus in epileptic patients is extremely tedious and time consuming. For this reason, many algorithms have been recently developed to detect HFOs. However, the developed algorithms suffer from false positives detection resulting from filtered-spikes without HFOs and sharp transients activities. HFOs exist in the frequency ban 80-50Hz and divided into Ripples 80-250 Hz and Fast Ripples 250-500Hz. The transient activities cover a wide bandwith from low to high frequencies and merely resemble HFOs events when filtered using classical band pass filters. However, using classical filtering methods based on FIR filters, wavelet transforms and the matching pursuit cannot separate the oscillatory from transient activities. This paper describes a practical approachof resonance based-filtering for decomposing intracranial EEG recordings of epileptic patients into the sum of an oscillatory component and transient component using over-complete rational dilation wavelet transforms (over complete RADWT) in conjunction with morphological component analysis (MCA).