Yi Sheng Zhu

Applications of adaptive filtering to ECG analysis: noise cancellation and arrhythmia detection

Several adaptive filter structures are proposed for noise cancellation and arrhythmia detection. The adaptive filter essentially minimizes the mean-squared error between a primary input, which is the noisy electrocardiogram (ECG), and a reference input, which is either noise that is correlated in some way with the noise in the primary input or a signal that is correlated only with ECG in the primary input. Different filter structures are presented to eliminate the diverse forms of noise: baseline wander, 60 Hz power line interference, muscle noise, and motion artifact. An adaptive recurrent filter structure is proposed for acquiring the impulse response of the normal QRS complex. The primary input of the filter is the ECG signal to be analyzed, while the reference input is an impulse train coincident with the QRS complexes. This method is applied to several arrhythmia detection problems: detection of P-waves, premature ventricular complexes, and recognition of conduction block, atrial fibrillation, and paced rhythm.<<ETX>>

Detecting ventricular tachycardia and fibrillation by complexity measure

Sinus rhythm (SR), ventricular tachycardia (VT) and ventricular fibrillation (VF) belong to different nonlinear physiological processes with different complexity. In this study, the authors present a novel, and computationally fast method to detect VT and VF, which utilizes a complexity measure suggested by Lempel and Ziv (1976). For a specific window length (i.e., the length of data segment to be analyzed), the method first generates a 0-1 string by comparing the raw electrocardiogram (ECG) data to a selected suitable threshold. The complexity measure can be obtained from the 0-1 string only using two simple operations, comparison and accumulation. When the window length is 7 s, the detection accuracy for each of SR, VT, and VF is 100% for a test set of 204 body surface records (34 SR, 85 monomorphic VT, and 85 VF). Compared with other conventional time- and frequency-domain methods, such as rate and irregularity, VF-filter leakage, and sequential hypothesis testing, the new algorithm is simple, computationally efficient, and well suited for real-time implementation in automatic external defibrillators (AEDs).

Ventricular tachycardia and fibrillation detection by a sequential hypothesis testing algorithm

An algorithm for detecting ventricular fibrillation (VF) and ventricular tachycardia (VT) by the method of sequential hypothesis testing is presented. The algorithm first generates a binary sequence by comparing the signal to a threshold. The probability distribution of the time intervals of the binary sequence is obtained, and the sequential hypothesis testing procedure of A.J. Wald and J. Wolfowitz (1948) is employed to discriminate the arrhythmias. Sequential hypothesis testing of 85 cases resulted in identification of (1) 97.64% VF and 97.65% VT episodes after 5 s and (2) 100% identification of both VF and VT after 7 s. The desired false positive and false negative error probabilities can be programmed into the algorithm. An important feature of the sequential method is that extra time for detection can be traded off for improved accuracy, and vice versa.<<ETX>>

A short-time multifractal approach for arrhythmia detection based on fuzzy neural network

The authors have proposed the notion of short-time multifractality and used it to develop a novel approach for arrhythmia detection. Cardiac rhythms are characterized by short-time generalized dimensions (STGDs), and different kinds of arrhythmias are discriminated using a neural network. To advance the accuracy of classification, a new fuzzy Kohonen network, which overcomes the shortcomings of the classical algorithm, is presented. In the authors paper, the potential of their method for clinical uses and real-time detection was examined using 180 electrocardiogram records [60 atrial fibrillation, 60 ventricular fibrillation, and 60 ventricular tachycardia]. The proposed algorithm has achieved high accuracy (more than 97%) and is computationally fast in detection.

Nonextensive entropy measure of EEG following brain injury from cardiac arrest

The nonextensive entropy measure is developed to study the electroencephalogram (EEG) during the recovery of the brains electrical function from asphyxic cardiac arrest (ACA) injury. The statistical characteristics of the Tsallis-like time-dependent entropy (TDE) for different signal distributions are investigated. Both the mean and the variance of TDE show good specificity to the ACA brain injury and its recovery. ACA brain injury results in a decrease in entropy while a good electrophysiological recovery shows a rapid return to a higher entropy level. There is a reduction in the mean and increase in the variance of TDE after brain injury followed by a gradual recovery upon resuscitation. The nonextensive TDE is expected to provide a novel quantitative EEG strategy for monitoring the brain states.

Removal of ECG interference from the EEG recordings in small animals using independent component analysis

In experiments involving small animals, the electroencephalogram (EEG) recorded during severe injury and accompanying resuscitation exhibit the strong presence of electrocardiogram (ECG). For improved quantitative EEG (qEEG) analysis, it is therefore imperative to remove ECG interference from EEG. In this paper, we validate the use of independent component analysis (ICA) to effectively suppress the interference of ECG from EEG recordings during normal activity, asphyxia and recovery following asphyxia. Two channels of EEG from five rats were recorded continuously for 2 h. Simultaneous recording of one channel ECG was also made. Epochs of 4 s and 1 min were selected from baseline, asphyxia and recovery (every 10 min) and their independent components and power spectra were calculated. The improvement in normalized power spectrum of EEG obtained for all animals was 7.71+/-3.63 db at the 3rd minute of recovery and dropped to 1.15+/-0.60 db at 63rd minute. The application of ICA has been particularly useful when the power of EEG is low, such as that observed during early brain hypoxic-asphyxic injury. The method is also useful in situations where accurate indications of EEG signal power and frequency content are needed.

Detection of non-linearity in the EEG of schizophrenic patients

OBJECTIVEnThe aim of this study is to detect non-linearity in the EEG of schizophrenia with a modified method of surrogate data. We also want to identify if dimension complexity (correlation dimension using spatial embedding) could be used as a discriminating statistic to demonstrate non-linearity in the EEG. The difference between the attractor dimension of healthy subjects and schizophrenic subjects is expected to be interpreted as reflecting some mechanisms underlying brain wave by views of non-linear dynamics analysis may reflect mechanistic differences.nnnMETHODSnEEGs were recorded with 14 electrodes in 18 healthy male subjects (average age: 26.3; range: 20--35) and 18 male schizophrenic patients (average age: 30.6; range: 24--40) during a resting eye-closed state. Neither of two groups was taking medicines. All artificial epochs in the EEG records were rejected by an experienced doctors visual inspection.nnnRESULTSnTesting non-linearity with modified surrogate data, we showed that correlation dimension of EEG data of schizophrenia does refuse the null hypothesis that the data were resulted from a linear dynamic system. A decrease of dimension complexity was found in the EEG of schizophrenia compared with controls. We interpreted it as the result of the psychopaths dysfunction overall brain. The surrogating procedure results in a significant increase in D(s).nnnCONCLUSIONSnNon-linearity of the EEG in schizophrenia was proven in our study. We think the correlation dimension with spatial embedding as a good discriminating statistic for testing such non-linearity. Moreover, schizophrenic patients EEGs were compared with controls and a lower dimension complexity was found. The results of our study indicate the possibility of using the methods of non-linear time series analysis to identify the EEGs of schizophrenic patients.

Modeling the relationship between concurrent epicardial action potentials and bipolar electrograms

A signal analysis approach to building the relationship between concurrent epicardial cell action potentials (APs) and bipolar electrograms is presented. Wavelet network, one nonlinear black-box modeling method, is used to identify the relationship between cell APs and bipolar electrocardiograms. The electrical signals were simultaneously measured from the epicardium of isolated Langendorff-perfused rabbit hearts during three different rhythm conditions: normal sinus rhythm (NSR), normal sinus rhythm after ischemia (NSRI), and ventricular fibrillation (VP). For NSR and NSRI, the proposed modeling method successfully captures the nonlinear input-output relationship and provides an accurate output, but the method fails in case of VF. This result suggests that a time-invariant nonlinear modeling method such as wavelet network is not appropriate for VF rhythm, which is thought to be time-varying as well as chaotic, but still useful in detection of VF. A new arrhythmia detection algorithm, with potential application in implantable devices, is proposed for identifying the time of rhythmic bifurcation.

Application of particle swarm system as a novel parameter optimization technique on spatiotemporal retina model

Center-surround spatiotemporal (ST) filter is a powerful tool to simulate the spatial and temporal properties of retina ganglion cells and encode visual information with electric spikes. This paper introduces the application of particle swarm optimization (PSO) algorithm to tune the parameters in the retina model consisting of a ST filter module and a back-propagation (BP) neural network module. Images are converted into electric spikes by the ST filters whose outputs are then fed into the BP neural network to reconstruct the output images. The parameters of the ST filters determine the electric spike sequences as well as the output image from the BP network. In order to get the expected output images, we employ PSO to iteratively tune the parameters. Euclidean distance between output and input image is used as scalar criteria to optimize the ST filter. The tuning process stops until the similarity between output and input images no longer improves. The results show that 62.3 % of the images trained by PSO have better output image quality and less iteration time compared with those trained by the current evolution strategy (ES).

Monitoring brain injury with Tsallis entropy

Nonextensive entropy measure, Tsallis entropy (TE), was undertaken to monitor the brain injury after cardiac arrest. EEG of human and experimental injury model of rats are investigated. In both conditions TE decreases in bad physiological functional outcome. As the brain recovers from injury, the TE will also gradually return to normal level. Meanwhile, TE also shows good sensitivity to different grades of asphyxic injury. This method provides a novel real time brain injury indicator and may be useful in developing a diagnostic monitoring tool.