Pei-Kuang Chao

Correlates of the shift in heart rate variability with postures and walking by time-frequency analysis

Heart rate (HR) variability derived from electrocardiogram (ECG) can be used to assess the function of the autonomic nervous system. HR exhibits various characteristics during different physical activities attributed to the altered autonomic mediation, where it is also beneficial to reveal the autonomic shift in response to physical-activity change. In this paper, the physical-activity-related HR behaviors were delineated using a portable ECG and body acceleration recorder based on a personal digital assistant and the smoothed pseudo Wigner-Ville distribution. The results based upon eighteen subjects performing four sequential 5-min physical activities (supine, sitting, standing and spontaneous walking) showed that the high-frequency heartbeat fluctuations during supine and sitting were significantly larger than during standing, and that the ratio of low- to high-frequency fluctuation during standing was significantly higher than during supine and sitting. This could be linked with the parasympathetic predominance during supine and sitting, and a shift to sympathetic dominance while standing. During spontaneous walking, the high-frequency fluctuation was significant lower than during supine. The low- to high-frequency ratio decreased significantly from standing to spontaneous walking, which may imply an increased vagal predominance (autonomic effect) or an increased respiratory activity (mechanical effect).

Detection of neuronal spikes using an adaptive threshold based on the max–min spread sorting method

Neuronal spike information can be used to correlate neuronal activity to various stimuli, to find target neural areas for deep brain stimulation, and to decode intended motor command for brain-machine interface. Typically, spike detection is performed based on the adaptive thresholds determined by running root-mean-square (RMS) value of the signal. Yet conventional detection methods are susceptible to threshold fluctuations caused by neuronal spike intensity. In the present study we propose a novel adaptive threshold based on the max-min spread sorting method. On the basis of microelectrode recording signals and simulated signals with Gaussian noises and colored noises, the novel method had the smallest threshold variations, and similar or better spike detection performance than either the RMS-based method or other improved methods. Moreover, the detection method described in this paper uses the reduced features of raw signal to determine the threshold, thereby giving a simple data manipulation that is beneficial for reducing the computational load when dealing with very large amounts of data (as multi-electrode recordings).

Classification of neuronal spikes over the reconstructed phase space.

Spike information is beneficial to correlate neuronal activity to various stimuli or determine target neural area for deep brain stimulation. Data clustering based on neuronal spike features provides a way to separate spikes generated from different neurons. Nevertheless, some spikes are aligned incorrectly due to spike deformation or noise interference, thereby reducing the accuracy of spike classification. In the present study, we proposed unsupervised spike classification over the reconstructed phase spaces of neuronal spikes in which the derived phase space portraits are less affected by alignment deviations. Principal component analysis was used to extract major principal components of the portrait features and k-means clustering was used to distribute neuronal spikes into various clusters. Finally, similar clusters were iteratively merged based upon inter-cluster portrait differences.

Unsupervised wavelet-based spike sorting with dynamic codebook searching and replenishment

Spike information is beneficial for correlating neuronal activity to various stimuli, finding target neural areas for deep brain stimulation, and decoding intended motor command for brain-machine interface. Unsupervised classification based on spike features provides a way to separate spikes generated from different neurons. Here, we propose an unsupervised spike sorting method based on specific wavelet coefficients (SWC) and using both a new spike alignment technique based on multi-peak energy comparison (MPEC) and a dynamic codebook-based template-matching algorithm with a class-merging feature. The MPEC alignment reduced inconsistent alignment caused by spike deformation. Using SWC not only reduced the number of features but also performed better in terms of matching a neuronal spike to its own class than relying on spike waveform or whole wavelet coefficients. Moreover, the employed codebook searching and replenishment can be operated in an online, real-time mode.

Wireless body area network for physical-activity classification and fall detection

A body area network constructed by a low-power microcontroller- (MSP430F149) and 2.4-GHz RF transceiver-(nRF24L01) is proposed to record body accelerations from multiple sites. The wireless setups reduce the inconvenience for free-moving due to the line connection between sensors and the central data processing recorder. The body area network was demonstrated useful in our experiments, while the subjects performed a series of scripted physical activity, free-moving, simulated falls and postural changes. In this system, a secure digital memory card was used to store measured acceleration signals in the central unit. The recorded data were processed to classify physical activity of the carrier by a multi-layer fuzzy clustering in a personal computer. Falls were detected by analyzing the changing magnitude of tri-axial accelerations. The proposed system can be applied to measure daily physical activity and detect falls in free-moving patients and the elderly.

Complex analysis of neuronal spike trains of deep brain nuclei in patients with Parkinson's disease.

Deep brain stimulation (DBS) of the subthalamic nucleus (STN) has been used to alleviate symptoms of Parkinsons disease. During image-guided stereotactic surgery, signals from microelectrode recordings are used to distinguish the STN from adjacent areas, particularly from the substantia nigra pars reticulata (SNr). Neuronal firing patterns based on interspike intervals (ISI) are commonly used. In the present study, arrival time-based measures, including Lempel-Ziv complexity and deviation-from-Poisson index were employed. Our results revealed significant differences in the arrival time-based measures among non-motor STN, motor STN and SNr and better discrimination than the ISI-based measures. The larger deviations from the Poisson process in the SNr implied less complex dynamics of neuronal discharges. If spike classification was not used, the arrival time-based measures still produced statistical differences among STN subdivisions and SNr, but the ISI-based measures only showed significant differences between motor and non-motor STN. Arrival time-based measures are less affected by spike misclassifications, and may be used as an adjunct for the identification of the STN during microelectrode targeting.

Recognition of Ventricular Extrasystoles Over the Reconstructed Phase Space of Electrocardiogram

Distinguishing ventricular extrasystoles from normal heartbeats is crucial to cardiac arrhythmia analysis. This paper proposes novel morphological descriptors, the major portrait partition area (MPPA) and point distribution percentage (PDP), which are extracted from the reconstructed phase space of the QRS complex. These measures can be linked to QRS width and prolonged ventricular contraction, and offer several advantages over traditional characterization of the QRS structure: it does not require QRS boundary detection, is robust under R-peak misalignment, and including some information from nearby points. The first four principal components of MPPA variables and PDPs in the first and the third quadrants of the phase space diagram were used as inputs of neural networks. The performance of networks in distinguishing premature ventricular contraction events from normal heartbeats were evaluated under a series of 50 cross-validations based on the electrocardiogram data taken from the MIT/BIH arrhythmia database. The sensitivity and specificity obtained using the aforementioned MPPA principal components and PDPs as inputs were similar to those obtained using wavelet features and Hermite coefficients. However, the phase space information performed better in situations of noise contaminations and waveform deformations.

An intelligent classifier for prognosis of cardiac resynchronization therapy based on speckle-tracking echocardiograms

PURPOSE Predicting response after cardiac resynchronization therapy (CRT) has been a challenge of cardiologists. About 30% of selected patients based on the standard selection criteria for CRT do not show response after receiving the treatment. This study is aimed to build an intelligent classifier to assist in identifying potential CRT responders by speckle-tracking radial strain based on echocardiograms. METHODS AND MATERIALS The echocardiograms analyzed were acquired before CRT from 26 patients who have received CRT. Sequential forward selection was performed on the parameters obtained by peak-strain timing and phase space reconstruction on speckle-tracking radial strain to find an optimal set of features for creating intelligent classifiers. Support vector machine (SVM) with a linear, quadratic, and polynominal kernel were tested to build classifiers to identify potential responders and non-responders for CRT by selected features. RESULTS Based on random sub-sampling validation, the best classification performance is correct rate about 95% with 96-97% sensitivity and 93-94% specificity achieved by applying SVM with a quadratic kernel on a set of 3 parameters. The selected 3 parameters contain both indexes extracted by peak-strain timing and phase space reconstruction. CONCLUSIONS An intelligent classifier with an averaged correct rate, sensitivity and specificity above 90% for assisting in identifying CRT responders is built by speckle-tracking radial strain. The classifier can be applied to provide objective suggestion for patient selection of CRT.

Phase-space Reconstruction of Electrocardiogram for Heartbeat Classification

Heartbeat classification is crucial for cardiac arrhythmia analysis. QRS complex presents important characteristics which are beneficial to distinguish abnormal beats from normal beats. In the present study we propose a novel descriptor for QRS complex. The waveform is transformed to a two-dimensional phase space and then mapped to a onedimensional portrait partition area (PPA). The proposed morphological descriptor has advantages of no need to detect Q and S characteristic points, tolerating R-peak misalignment and taking into account temporal relation of data samples. On the basis of 32 records from the MIT/BIH arrhythmia database, normal QRS and premature ventricular contraction (PVC) beats show different phase space portraits and PPA. An artificial neuronal network using PPA as the input feature was built for heartbeat classification. Our results showed that the sensitivity and specificity of distinguishing PVC from normal QRS achieved 0.9699 and 0.9651 in the testing sets, respectively.

Brain connectivity of patients with Alzheimer's disease by coherence and cross mutual information of electroencephalograms during photic stimulation.

Alzheimers disease (AD) is a neurodegenerative disease, usually diagnosed by neuropsychological tests, and excluded from other cerebral diseases by brain images. An electroencephalogram (EEG) provides a means of disclosing the reduced functional couplings between brain regions that occurs with AD. In the present study, 16 probable AD patients and 15 age-matched, gender-matched normal subjects were enrolled. Spectral coherence and cross mutual information (CMI) were used to analyze EEGs during intermittent photic stimulation (PS). Ocular- and heartbeat-related source components (SCs) obtained from multi-channel EEGs by the independent component analysis were discarded, and the photic-related SCs were reduced using a comb filter. The undisturbed SCs and photic-related SCs before and after photic reduction were used to reconstruct photic-preserved EEGs and photic-reduced EEGs, from which harmonic coherences (direct photic-driving response) and rhythmic coherences and CMI (indirect photic affection) were computed, respectively. Our results indicate that the rhythmic coherences (particularly in the alpha and beta bands) and CMI variables as well as the harmonic coherences (particularly related to 3-Hz PS) were significantly lower in the probable AD than in normal subjects, whereas the variables derived from the resting EEGs were not statistically significant. This finding implied that the variables obtained during PS could be used to disclose impaired intra-brain associations in probable AD.