This paper proposes non-Gaussian models for parametric spectral estimation with application to event-related desynchronization (ERD) estimation of nonstationary EEG. Existing approaches for time-varying spectral estimation use time-varying autoregressive (TVAR) state-space models with Gaussian state noise. The parameter estimation is solved by a conventional Kalman filtering. This study uses non-Gaussian state noise to model autoregressive (AR) parameter variation with estimation by a Monte Carlo particle filter (PF). Use of non-Gaussian noise such as heavy-tailed distribution is motivated by its ability to track abrupt and smooth AR parameter changes, which are inadequately modeled by Gaussian models. Thus, more accurate spectral estimates and better ERD tracking can be obtained. This study further proposes a non-Gaussian state space formulation of time-varying autoregressive moving average (TVARMA) models to improve the spectral estimation. Simulation on TVAR process with abrupt parameter variation shows superior tracking performance of non-Gaussian models. Evaluation on motor-imagery EEG data shows that the non-Gaussian models provide more accurate detection of abrupt changes in alpha rhythm ERD. Among the proposed non-Gaussian models, TVARMA shows better spectral representations while maintaining reasonable good ERD tracking performance.
A signal subspace approach for extracting visual evoked potentials (VEPs) from the background electroencephalogram (EEG) colored noise without the need for a prewhitening stage is proposed. Linear estimation of the clean signal is performed by minimizing signal distortion while maintaining the residual noise energy below some given threshold. The generalized eigendecomposition of the covariance matrices of a VEP signal and brain background EEG noise is used to transform them jointly to diagonal matrices. The generalized subspace is then decomposed into signal subspace and noise subspace. Enhancement is performed by nulling the components in the noise subspace and retaining the components in the signal subspace. The performance of the proposed algorithm is tested with simulated and real data, and compared with the recently proposed signal subspace techniques. With the simulated data, the algorithms are used to estimate the latencies of P(100), P(200), and P(300) of VEP signals corrupted by additive colored noise at different values of SNR. With the real data, the VEP signals are collected at Selayang Hospital, Kuala Lumpur, Malaysia, and the capability of the proposed algorithm in detecting the latency of P(100) is obtained and compared with other subspace techniques. The ensemble averaging technique is used as a baseline for this comparison. The results indicated significant improvement by the proposed technique in terms of better accuracy and less failure rate.
In this paper, we investigate the use of heart rate variability (HRV) for automatic newborn seizure detection. The proposed method consists of a sequence of processing steps, namely, obtaining HRV from the ECG, extracting a discriminating HRV feature set, selecting an optimal subset from the full feature set, and, finally, classifying the HRV into seizure/nonseizure using a supervised statistical classifier. Due to the fact that HRV signals are nonstationary, a set of time-frequency features from the newborn HRV is proposed and extracted. In order to achieve efficient HRV-based automatic newborn seizure detection, a two-phase wrapper-based feature selection technique is used to select the feature subset with minimum redundancy and maximum class discriminability. Tested on ECG recordings obtained from eight newborns with identified EEG seizure, the proposed HRV-based neonatal seizure detection algorithm achieved 85.7% sensitivity and 84.6% specificity. These results suggest that the HRV is sensitive to changes in the cardioregulatory system induced by the seizure, and therefore, can be used as a basis for an automatic seizure detection.
The mu rhythm is an electroencephalogram (EEG) signal located at the central region of the brain that is frequently used for studies concerning motor activity. Quite often, the EEG data are contaminated with artifacts and the application of blind source separation (BSS) alone is insufficient to extract the mu rhythm component. We present a new two-stage approach to extract the mu rhythm component. The first stage uses second-order blind identification (SOBI) with stationary wavelet transform (SWT) to automatically remove the artifacts. In the second stage, SOBI is applied again to find the mu rhythm component. Our method is first compared with independent component analysis with discrete wavelet transform (ICA-DWT) as well as SOBI-DWT, ICA-SWT, and regression method for artifact removal using simulated EEG data. The results showed that the regression method is more effective in removing electrooculogram (EOG) artifacts, while SOBI-SWT is more effective in removing electromyogram (EMG) artifacts as compared to the other artifact removal methods. Then, all the methods are compared with the direct application of SOBI in extracting mu rhythm components on simulated and actual EEG data from ten subjects. The results showed that the proposed method of SOBI-SWT artifact removal enhances the extraction of the mu rhythm component.
In obstetrics, fetal heart rate (FHR) detection remains the standard for intrapartum assessment of fetal well-being. In this paper, a low-power (< 55 mW) optical technique is proposed for transabdominal FHR detection using near-infrared photoplesthysmography (PPG). A beam of IR-LED (890 nm) propagates through to the maternal abdomen and fetal tissues, resulting in a mixed signal detected by a low-noise detector situated at a distance of 4 cm. Low-noise amplification and 24-bit analog-to-digital converter resolution ensure minimum effect of quantization noise. After synchronous detection, the mixed signal is processed by an adaptive filter to extract the fetal signal, whereas the PPG from the mother's index finger is the reference input. A total of 24 datasets were acquired from six subjects at 37 +/- 2 gestational weeks. Results show a correlation coefficient of 0.96 (p-value < 0.001) between the proposed optical and ultrasound FHR, with a maximum error of 4%. Assessment of the effect of probe position on detection accuracy indicates that the probe should be close to fetal tissues, but not necessarily restricted to head or buttocks.
An algorithm based on digital filtering, adaptive thresholding, statistical properties in the time domain, and differencing of local maxima and minima has been developed for the simultaneous measurement of the fetal and maternal heart rates from the maternal abdominal electrocardiogram during pregnancy and labor for ambulatory monitoring. A microcontroller-based system has been used to implement the algorithm in real-time. A Doppler ultrasound fetal monitor was used for statistical comparison on five volunteers with low risk pregnancies, between 35 and 40 weeks of gestation. Results showed an average percent root mean square difference of 5.32% and linear correlation coefficient from 0.84 to 0.93. The fetal heart rate curves remained inside a +/- 5-beats-per-minute limit relative to the reference ultrasound method for 84.1% of the time.