Detecting ventricular tachycardia and fibrillation by complexity measure

Machine Learning Techniques for the Detection of Shockable Rhythms in Automated External Defibrillators

Early recognition of ventricular fibrillation (VF) and electrical therapy are key for the survival of out-of-hospital cardiac arrest (OHCA) patients treated with automated external defibrillators (AED). AED algorithms for VF-detection are customarily assessed using Holter recordings from public electrocardiogram (ECG) databases, which may be different from the ECG seen during OHCA events. This study evaluates VF-detection using data from both OHCA patients and public Holter recordings. ECG-segments of 4-s and 8-s duration were analyzed. For each segment 30 features were computed and fed to state of the art machine learning (ML) algorithms. ML-algorithms with built-in feature selection capabilities were used to determine the optimal feature subsets for both databases. Patient-wise bootstrap techniques were used to evaluate algorithm performance in terms of sensitivity (Se), specificity (Sp) and balanced error rate (BER). Performance was significantly better for public data with a mean Se of 96.6%, Sp of 98.8% and BER 2.2% compared to a mean Se of 94.7%, Sp of 96.5% and BER 4.4% for OHCA data. OHCA data required two times more features than the data from public databases for an accurate detection (6 vs 3). No significant differences in performance were found for different segment lengths, the BER differences were below 0.5-points in all cases. Our results show that VF-detection is more challenging for OHCA data than for data from public databases, and that accurate VF-detection is possible with segments as short as 4-s.

Wavelet Based Method for Congestive Heart Failure Recognition by Three Confirmation Functions

An investigation of the electrocardiogram (ECG) signals and arrhythmia characterization by wavelet energy is proposed. This study employs a wavelet based feature extraction method for congestive heart failure (CHF) obtained from the percentage energy (PE) of terminal wavelet packet transform (WPT) subsignals. In addition, the average framing percentage energy (AFE) technique is proposed, termed WAFE. A new classification method is introduced by three confirmation functions. The confirmation methods are based on three concepts: percentage root mean square difference error (PRD), logarithmic difference signal ratio (LDSR), and correlation coefficient (CC). The proposed method showed to be a potential effective discriminator in recognizing such clinical syndrome. ECG signals taken from MIT-BIH arrhythmia dataset and other databases are utilized to analyze different arrhythmias and normal ECGs. Several known methods were studied for comparison. The best recognition rate selection obtained was for WAFE. The recognition performance was accomplished as 92.60% accurate. The Receiver Operating Characteristic curve as a common tool for evaluating the diagnostic accuracy was illustrated, which indicated that the tests are reliable. The performance of the presented system was investigated in additive white Gaussian noise (AWGN) environment, where the recognition rate was 81.48% for 5 dB.

Effects of religious and spiritual variables on outcomes in violent relationships

OBJECTIVE

Religious and spiritual factors in intimate partner violence have received increasing attention. But are such factors related to outcomes in violent relationships? The purpose of this study was to assess the relative impact of spiritual symptoms and religious coping on attitudinal/behavioral and clinical outcomes among women in violent relationships.

METHODS

Adult women with a recent history of husband-to-wife physical abuse were recruited from six primary care clinics. Once enrolled, 200 subjects completed a baseline interview and daily assessment of level of violence, using the Interactive Verbal Response for 12 weeks. At the completion of the study, contact with each participant was attempted to determine whether she had either sought professional help or left the relationship. Three religious/spiritual variables were assessed at baseline-number of visits to a religious/spiritual counselor, religious coping, and severity of spiritual symptoms. Stepped multiple linear regression was used to explain factor-analyzed outcomes (coping and appraisals, hope and support, symptomatology, functional status, readiness for change, and medical utilization), adjusting for demographic, marital, childhood, mental health, and violence variables.

RESULTS

After controlling for duration, severity and dynamics of violence, the use of spiritual resources, and the level of spiritual symptoms were associated with most attitudinal/behavioral and clinical outcomes, while religious coping was only associated with staying in the relationship.

CONCLUSIONS

Religious and spiritual factors were associated with most outcomes. Spiritual symptoms had a consistently negative effect on outcomes while use of spiritual resources had variable effects. Religious coping was only associated with refraining from leaving the relationship.

Multi-day recurrences of intimate partner violence and alcohol intake across dynamic patterns of violence

RATIONALE, AIMS AND OBJECTIVES

Consistent links exist between male alcohol intake and male-perpetrated intimate partner violence (IPV) as well as female alcohol intake and female-perpetrated IPV. However, the nature of the relationship remains unclear. This study attempted to identify unique alcohol-violence patterns within three different types of relationship dynamics to better understand the alcohol-violence relationship and its role in violence dynamics.

METHOD

Two hundred women in abusive relationships were recruited from six primary care clinics. Subjects completed daily assessments of their relationship using interactive verbal response via telephone for 12 weeks. Dynamic patterns (periodic, chaotic, random) were determined by positive versus negative Lyapunov exponents and measures of correlation dimension saturation. To identify recurrent day-to-day activities, we used orbital decomposition (based on symbolic dynamics).

RESULTS

Periodic dynamics included daily reports with mutual abuse and alcohol intake while random dynamics included a variety of patterns, especially those involving unequal mutual abuse. Unique strings for each dynamic pattern were examined. Periodic dynamics involved heavy alcohol intake by the husband or mutual moderate-severe violence. Random dynamics uniquely involved mutual verbal abuse with husband's alcohol intake on same or different days as well as husband-perpetrated moderate-severe violence with or without husband-perpetrated minor violence. Chaotic dynamics uniquely involved combinations from wife-perpetrated minor violence alone to combinations of husband's heavy alcohol intake (with or without husband-perpetrated minor violence), mutual verbal abuse, and husband-perpetrated verbal abuse (with or without husband's heavy alcohol intake).

CONCLUSION

Recurrent 4-day patterns were observed. Each dynamic pattern was characterized by recurrent strings unique to that pattern.

Do violence dynamics matter?

RATIONALE, AIMS AND OBJECTIVES

Intimate partner violence is a complex, non-linear phenomenon. The purpose of this study was to determine whether violence dynamics (pattern, degree of non-linearity, optimal non-linearity) contributed to outcomes in violent relationships.

METHODS

The study was conducted in six primary care clinics, enrolling 200 adult women in violent relationships. In addition to baseline and end-of-study interviews, women completed daily telephone assessments of household environment and partner violence using interactive verbal response. Three non-linearity measures of violence were computed with 'optimal' non-linearity estimated using Z-transformations. Assignment of dynamic patterns (periodic, chaotic, random) was made based upon Lyapunov exponent and correlation dimension. Outcomes across dynamic patterns were analysed using analysis of variance. In addition, stepped multiple linear regression explained factor-analysed outcomes, adjusting for demographic, childhood, mental health and marital variables; attitudinal/behavioural outcomes were also adjusted for when explaining clinical outcomes.

RESULTS

Women experiencing periodic violence recognized the importance of violence and used their active coping to seek mental health care. Those with chaotic dynamics recognized that they were not responsible, experienced fewer psychological symptoms and emotional role limitations, and did not seek help. Those experiencing random violence recognized its unpredictability and uncontrollability. Violence non-linearity predicted negative coping, positive appraisals and hope/support in regression analyses, while optimal non-linearity contributed to readiness for change and symptoms functioning. Of the nine outcomes investigated, violence non-linearity contributed to five outcomes.

CONCLUSION

Dynamic pattern of violence, degree of violence non-linearity and optimal non-linearity correlated with several attitudinal/behavioural and clinical outcomes. Knowledge of violence dynamics may have applications when working with violent couples.

Dynamics of violence

RATIONALE, AIMS AND OBJECTIVES

Three behavioural models suggest different dynamic patterns of intimate partner violence (IPV). However, few studies permit assessment of IPV dynamics. The purpose of this study was to estimate the degree of non-linearity in daily violence between partners over a 3-month period, identify their specific dynamic patterns and determine whether measures of violence severity and dynamics are interrelated.

METHODS

From six primary care clinics, we enrolled 200 adult women who experienced violence in the previous month and asked them to complete daily telephone assessments of household environment, marital relationship and violence using Interactive Verbal Response. To assess non-linearity of violence, algorithmic complexity was measured by LZ complexity and lack of regularity was measured by approximate entropy. Lyapunov exponents and correlation dimension saturation were used to approximate dynamic patterns.

RESULTS

Of the 9618 daily reports, women reported experiencing abuse on 39% of days, while perpetrating violence themselves on 23% of days. Most (59%) displayed random dynamics, 30% showed chaotic and 12% showed periodic dynamics. All three measures of non-linearity consistently demonstrated non-linear patterns of violence. Using multivariate analysis of variance, neither episode severity for men or women showed significant differences across dynamic types, but chaotic dynamics had the lowest frequencies of violence in men and women while random dynamics had the highest frequencies. Approximate entropy was positively correlated with violence frequency and burden in men and women, but Lyapunov exponent was inversely related to violence. LZ complexity correlated positively with wife-perpetrated violence only.

CONCLUSIONS

IPV is rarely a predictable, periodic phenomenon; no behavioural model describes the violence dynamics for all violent relationships. Yet, the measures of non-linearity and specific dynamic patterns correlate with different violent features of these relationships.

Electrocardiogram pattern recognition and analysis based on artificial neural networks and support vector machines: a review

Computer systems for Electrocardiogram (ECG) analysis support the clinician in tedious tasks (e.g., Holter ECG monitored in Intensive Care Units) or in prompt detection of dangerous events (e.g., ventricular fibrillation). Together with clinical applications (arrhythmia detection and heart rate variability analysis), ECG is currently being investigated in biometrics (human identification), an emerging area receiving increasing attention. Methodologies for clinical applications can have both differences and similarities with respect to biometrics. This paper reviews methods of ECG processing from a pattern recognition perspective. In particular, we focus on features commonly used for heartbeat classification. Considering the vast literature in the field and the limited space of this review, we dedicated a detailed discussion only to a few classifiers (Artificial Neural Networks and Support Vector Machines) because of their popularity; however, other techniques such as Hidden Markov Models and Kalman Filtering will be also mentioned.

Effects of the series length on Lempel-Ziv Complexity during sleep

Lempel-Ziv Complexity (LZC) has been demonstrated to be a powerful complexity measure in several biomedical applications. During sleep, it is still not clear how many samples are required to ensure robustness of its estimate when computed on beat-to-beat interval series (RR). The aims of this study were: i) evaluation of the number of necessary samples in different sleep stages for a reliable estimation of LZC; ii) evaluation of the LZC when considering inter-subject variability; and iii) comparison between LZC and Sample Entropy (SampEn). Both synthetic and real data were employed. In particular, synthetic RR signals were generated by means of AR models fitted on real data. The minimum number of samples required by LZC for having no changes in its average value, for both NREM and REM sleep periods, was 10(4) (p<;0.01) when using a binary quantization. However, LZC can be computed with N >1000 when a tolerance of 5% is considered satisfying. The influence of the inter-subject variability on the LZC was first assessed on model generated data confirming what found (>10(4); p<;0.01) for both NREM and REM stage. However, on real data, without differentiate between sleep stages, the minimum number of samples required was 1.8×10(4). The linear correlation between LZC and SampEn was computed on a synthetic dataset. We obtained a correlation higher than 0.75 (p<;0.01) when considering sleep stages separately, and higher than 0.90 (p<;0.01) when stages were not differentiated. Summarizing, we suggest to use LZC with the binary quantization and at least 1000 samples when a variation smaller than 5% is considered satisfying, or at least 10(4) for maximal accuracy. The use of more than 2 levels of quantization is not recommended.

Detection of life-threatening arrhythmias using feature selection and support vector machines

Early detection of ventricular fibrillation (VF) and rapid ventricular tachycardia (VT) is crucial for the success of the defibrillation therapy. A wide variety of detection algorithms have been proposed based on temporal, spectral, or complexity parameters extracted from the ECG. However, these algorithms are mostly constructed by considering each parameter individually. In this study, we present a novel life-threatening arrhythmias detection algorithm that combines a number of previously proposed ECG parameters by using support vector machines classifiers. A total of 13 parameters were computed accounting for temporal (morphological), spectral, and complexity features of the ECG signal. A filter-type feature selection (FS) procedure was proposed to analyze the relevance of the computed parameters and how they affect the detection performance. The proposed methodology was evaluated in two different binary detection scenarios: shockable (FV plus VT) versus nonshockable arrhythmias, and VF versus nonVF rhythms, using the information contained in the medical imaging technology database, the Creighton University ventricular tachycardia database, and the ventricular arrhythmia database. sensitivity (SE) and specificity (SP) analysis on the out of sample test data showed values of SE=95%, SP=99%, and SE=92% , SP=97% in the case of shockable and VF scenarios, respectively. Our algorithm was benchmarked against individual detection schemes, significantly improving their performance. Our results demonstrate that the combination of ECG parameters using statistical learning algorithms improves the efficiency for the detection of life-threatening arrhythmias.

Visual data mining with self-organising maps for ventricular fibrillation analysis

Detection of ventricular fibrillation (VF) at an early stage is being deeply studied in order to lower the risk of sudden death and allows the specialist to have greater reaction time to give the patient a good recovering therapy. Some works are focusing on detecting VF based on numerical analysis of time-frequency distributions, but in general the methods used do not provide insight into the problem. However, this study proposes a new methodology in order to obtain information about this problem. This work uses a supervised self-organising map (SOM) to obtain visually information among four important groups of patients: VF (ventricular fibrillation), VT (ventricular tachycardia), HP (healthy patients) and AHR (other anomalous heart rates and noise). A total number of 27 variables were obtained from continuous surface ECG recordings in standard databases (MIT and AHA), providing information in the time, frequency, and time-frequency domains. self-organising maps (SOMs), trained with 11 of the 27 variables, were used to extract knowledge about the variable values for each group of patients. Results show that the SOM technique allows to determine the profile of each group of patients, assisting in gaining a deeper understanding of this clinical problem. Additionally, information about the most relevant variables is given by the SOM analysis.

Ventricular fibrillation and tachycardia classification using a machine learning approach

Correct detection and classification of ventricular fibrillation (VF) and rapid ventricular tachycardia (VT) is of pivotal importance for an automatic external defibrillator and patient monitoring. In this paper, a VF/VT classification algorithm using a machine learning method, a support vector machine, is proposed. A total of 14 metrics were extracted from a specific window length of the electrocardiogram (ECG). A genetic algorithm was then used to select the optimal variable combinations. Three annotated public domain ECG databases (the American Heart Association Database, the Creighton University Ventricular Tachyarrhythmia Database, and the MIT-BIH Malignant Ventricular Arrhythmia Database) were used as training, test, and validation datasets. Different window sizes, varying from 1 to 10 s were tested. An accuracy (Ac) of 98.1%, sensitivity (Se) of 98.4%, and specificity (Sp) of 98.0% were obtained on the in-sample training data with 5 s-window size and two selected metrics. On the out-of-sample validation data, an Ac of 96.3% ± 3.4%, Se of 96.2% ± 2.7%, and Sp of 96.2% ± 4.6% were obtained by fivefold cross validation. The results surpass those of current reported methods.

Analysis of Intracranial Pressure

The monitoring of intracranial pressure (ICP) is an important tool in medicine for its ability to portray the brain’s compliance status. The bedside monitor displays the ICP waveform and intermittent mean values to guide physicians in the management of patients, particularly those having sustained a traumatic brain injury. Researchers in the fields of engineering and physics have investigated various mathematical analysis techniques applicable to the waveform in order to extract additional diagnostic and prognostic information, although they largely remain limited to research applications. The purpose of this review is to present the current techniques used to monitor and interpret ICP and explore the potential of using advanced mathematical techniques to provide information about system perturbations from states of homeostasis. We discuss the limits of each proposed technique and we propose that nonlinear analysis could be a reliable approach to describe ICP signals over time, with the fractal dimension as a potential predictive clinically meaningful biomarker. Our goal is to stimulate translational research that can move modern analysis of ICP using these techniques into widespread practical use, and to investigate to the clinical utility of a tool capable of simplifying multiple variables obtained from various sensors.

Interpretation of coarse-graining of Lempel-Ziv complexity measure in ECG signal analysis

Lempel-Ziv (LZ) complexity measure has been applied to classify ventricular tachycardia (VT) and ventricular fibrillation (VF). The coarse-graining process plays a crucial role in the LZ complexity measure analysis, which directly affects the separating performance of VT and VF in ECG signal analysis. The question of different coarse-graining approaches interpretability in ECG signal analysis and their influence on the performance of ECG classification have not yet been previously addressed in the literature. In this paper, we present four coarse-graining process approaches, K-Means, Mean, Median and Mid-point. Our test shows that K-Means algorithm is superior to the other three approaches in VT and VF separation rate, Particularly, optimum performance is achieved at a 8-second window length.

Brain oscillatory complexity across the life span

OBJECTIVE

Considering the increasing use of complexity estimates in neuropsychiatric populations, a normative study is critical to define the 'normal' behaviour of brain oscillatory complexity across the life span.

METHOD

This study examines changes in resting-state magnetoencephalogram (MEG) complexity - quantified with the Lempel-Ziv complexity (LZC) algorithm - due to age and gender in a large sample of 222 (100 males/122 females) healthy participants with ages ranging from 7 to 84 years.

RESULTS

A significant quadratic (curvilinear) relationship (p<0.05) between age and complexity was found, with LZC maxima being reached by the sixth decade of life. Once that peak was crossed, complexity values slowly decreased until late senescence. Females exhibited higher LZC values than males, with significant differences in the anterior, central and posterior regions (p<0.05).

CONCLUSIONS

These results suggest that the evolution of brain oscillatory complexity across the life span might be considered a new illustration of a 'normal' physiological rhythm.

SIGNIFICANCE

Previous and forthcoming clinical studies using complexity estimates might be interpreted from a more complete and dynamical perspective. Pathologies not only cause an 'abnormal' increase or decrease of complexity values but they actually 'break' the 'normal' pattern of oscillatory complexity evolution as a function of age.

Pulse amplitude and Lempel-Ziv complexity of the cerebrospinal fluid pressure signal

BACKGROUND

The complexity of the intracranial pressure (ICP) signal decreases with intracranial hypertension in children with acute brain injury as well as during infusion studies in adults with hydrocephalus. In this study we have analysed the pressure signal obtained in the lumbar subarachnoid space during infusion testing. The pulse amplitude rises when the ICP is increased by additional external volume. Our objective was to determine the relative influence of the pressure range and pulse amplitude on the loss of complexity observed during infusion-related intracranial hypertension.

MATERIALS AND METHODS

The Lempel-Ziv (LZ) complexity of the cerebrospinal fluid pressure (CSFP) signal was analysed in 52 infusion studies performed in patients with normal pressure hydrocephalus (median age 71 years, IQR: 60-78). Four sequences during the baseline, infusion, steady plateau and recovery periods of each infusion study were selected. The mean values of the CSFP (mCSFP), pulse amplitude and LZ complexity in every sequence were measured. Correlations between LZ complexity and CSFP parameters were explored.

RESULTS

Significant inverse correlations were found among LZ complexity, pulse amplitude and mCSFP during all periods of infusion testing, except at baseline. Partial correlation analysis controlling the effect of mCSFP emphasised the relationship between pulse amplitude and LZ complexity. When pulse amplitude is held constant the partial correlation between LZ complexity and mCSFP is not significant.

CONCLUSIONS

The pulse amplitude of the CSFP signal seems to be a major determinant of the waveform complexity.

Feature extraction for ECG heartbeats using higher order statistics of WPD coefficients

This paper describes feature extraction methods using higher order statistics (HOS) of wavelet packet decomposition (WPD) coefficients for the purpose of automatic heartbeat recognition. The method consists of three stages. First, the wavelet package coefficients (WPC) are calculated for each different type of ECG beat. Then, higher order statistics of WPC are derived. Finally, the obtained feature set is used as input to a classifier, which is based on k-NN algorithm. The MIT-BIH arrhythmia database is used to obtain the ECG records used in this study. All heartbeats in the arrhythmia database are grouped into five main heartbeat classes. The classification accuracy of the proposed system is measured by average sensitivity of 90%, average selectivity of 92% and average specificity of 98%. The results show that HOS of WPC as features are highly discriminative for the classification of different arrhythmic ECG beats.

Time-domain ECG signal analysis based on smart-phone

In this paper, a time domain algorithm architecture is presented and implemented on a smart-phone for ECG signal analysis. Using the QRS detection algorithm suggested by Pan-Tompkins and the beat classification method, the heart beats are detected and classified as normal beats and premature ventricular contractions (PVCs). Subsequently, a computationally efficient method is presented to separate ventricular tachycardia (VT) and ventricular fibrillation (VF). This method utilizes Lempel and Ziv complexity analysis combined with K-means algorithm for the coarse-graining process. In addition, a new classification rule is presented to recognize VT and VF in our study. The proposed system provides fairly good performance when applied to the MIT-BIH Database. This algorithm architecture can be efficiently used on the mobile platform.

ARTIFICIAL NEURAL NETWORK BASED ECG ARRHYTHMIA CLASSIFICATION

Reliable and computationally efficient means of classifying electrocardiogram (ECG) signals has been the subject of considerable research effort in recent years. This paper explores the potential applications of a talented, versatile computation model called the Artificial Neural Network (ANN) in the field of ECG signal classification. Two types of ANNs: Multi-Layered Feed Forward Network (MLFFN) and Probabilistic Neural Networks (PNN) are used to classify seven types of ECG beats. It includes six types of arrhythmia data and normal data. Here, parametric modeling strategies are used in conjunction with ANN classifiers to discriminate ECG signals. Instead of giving the ECG data as such, parameters such as fourth order Auto Regressive model coefficients and Spectral Entropy of the signals has been selected. On testing with the Massachusetts Institute of Technology-Beth Israel Hospital (MIT/BIH) arrhythmia database, it has been observed that PNN has better performance than conventionally used MLFFN in ECG arrhythmia classification. MLFFN with Back Propagation Algorithm gives a classification accuracy of 97.54% and PNN gives 98.96%. The classification by PNN also has an advantage that the computation time for classification is lower than that of MLFFN.

Life-threatening arrhythmia verification in ICU patients using the joint cardiovascular dynamical model and a Bayesian filter

In this paper, a novel nonlinear joint dynamical model is presented, which is based on a set of coupled ordinary differential equations of motion and a Gaussian mixture model representation of pulsatile cardiovascular (CV) signals. In the proposed framework, the joint interdependences of CV signals are incorporated by assuming a unique angular frequency that controls the limit cycle of the heart rate. Moreover, the time consequence of CV signals is controlled by the same phase parameter that results in the space dimensionality reduction. These joint equations together with linear assignments to observation are further used in the Kalman filter structure for estimation and tracking. Moreover, we propose a measure of signal fidelity by monitoring the covariance matrix of the innovation signals throughout the filtering procedure. Five categories of life-threatening arrhythmias were verified by simultaneously tracking the signal fidelity and the polar representation of the CV signal estimations. We analyzed data from Physiobank multiparameter databases (MIMIC I and II). Performance evaluation results demonstrated that the sensitivity of the detection ranges over 93.50% and 100.00%. In particular, the addition of more CV signals improved the positive predictivity of the proposed method to 99.27% for the total arrhythmic types. The method was also used for false arrhythmia suppression issued by ICU monitors, with an overall false suppression rate reduced from 42.3% to 9.9%. In addition, false critical ECG arrhythmia alarm rates were found to be, on average, 42.3%, with individual rates varying between 16.7% and 86.5%. The results illustrate that the method can contribute to, and enhance the performance of clinical life-threatening arrhythmia detection.

Complexity of VTA DA neural activities in response to PFC transection in nicotine treated rats

Background

The dopaminergic (DA) neurons in the ventral tegmental area (VTA) are widely implicated in the addiction and natural reward circuitry of the brain. These neurons project to several areas of the brain, including prefrontal cortex (PFC), nucleus accubens (NAc) and amygdala. The functional coupling between PFC and VTA has been demonstrated, but little is known about how PFC mediates nicotinic modulation in VTA DA neurons. The objectives of this study were to investigate the effect of acute nicotine exposure on the VTA DA neuronal firing and to understand how the disruption of communication from PFC affects the firing patterns of VTA DA neurons.

Methods

Extracellular single-unit recordings were performed on Sprague-Dawley rats and nicotine was administered after stable recording was established as baseline. In order to test how input from PFC affects the VTA DA neuronal firing, bilateral transections were made immediate caudal to PFC to mechanically delete the interaction between VTA and PFC.

Results

The complexity of the recorded neural firing was subsequently assessed using a method based on the Lempel-Ziv estimator. The results were compared with those obtained when computing the entropy of neural firing. Exposure to nicotine triggered a significant increase in VTA DA neurons firing complexity when communication between PFC and VTA was present, while transection obliterated the effect of nicotine. Similar results were obtained when entropy values were estimated.

Conclusions

Our findings suggest that PFC plays a vital role in mediating VTA activity. We speculate that increased firing complexity with acute nicotine administration in PFC intact subjects is due to the close functional coupling between PFC and VTA. This hypothesis is supported by the fact that deletion of PFC results in minor alterations of VTA DA neural firing when nicotine is acutely administered.

An algorithm used for ventricular fibrillation detection without interrupting chest compression

Ventricular fibrillation (VF) is the primary arrhythmic event in the majority of patients suffering from sudden cardiac arrest. Attention has been focused on this particular rhythm since it is recognized that prompt therapy, especially electrical defibrillation, may lead to a successful outcome. However, current versions of automated external defibrillators (AEDs) mandate repetitive interruptions of chest compression for rhythm analyses since artifacts produced by chest compression during cardiopulmonary resuscitation (CPR) preclude reliable electrocardiographic (ECG) rhythm analysis. Yet, repetitive interruptions in chest compression are detrimental to the success of defibrillation. The capability for rhythm analysis without requiring "hands-off" intervals will allow for more effective resuscitation. In this paper, a novel continuous-wavelet-transformation-based morphology consistency evaluation algorithm was developed for the detection of disorganized VF from organized sinus rhythm (SR) without interrupting the ongoing chest compression. The performance of this method was evaluated on both uncorrupted and corrupted ECG signals recorded from AEDs obtained from out-of-hospital victims of cardiac arrest. A total of 232 patients and 31,092 episodes of either VF or SR were accessed, in which 8195 episodes were corrupted by artifacts produced by chest compressions. We also compared the performance of this method with three other established algorithms, including VF filter, spectrum analysis, and complexity measurement. Even though there was a modest decrease in specificity and accuracy when chest compression artifact was present, the performance of this method was still superior to other reported methods for VF detection during uninterrupted CPR.

A multi-stage automatic arrhythmia recognition and classification system

This paper describes an automatic classification system based on combination of diverse features for the purpose of automatic heartbeat recognition. The method consists of three stages. At the first stage, heartbeats are classified into 5 main groups defined by AAMI using optimal feature sets for each main group. At the second stage, main groups are classified into subgroups using optimal features for each subgroup. Then the third stage is added to the system for classifying beats that are labeled as unclassified beats in the first two classification stages. A diverse set of features including higher order statistics, morphological features, Fourier transform coefficients, and higher order statistics of the wavelet package coefficients are extracted for each different type of ECG beat. At the first stage, optimal features for main groups are determined by using a wrapper type feature selection algorithm. At the second stage, optimal features are similarly selected for discriminating each subgroup of the main groups. Then at the third stage, only raw data is used for classifying beats. In all stages, the classifiers are based on the k-nearest neighbor algorithm. ECG records used in this study are obtained from the MIT-BIH arrhythmia database. The classification accuracy of the proposed system is measured by sensitivity, selectivity, and specificity measures. The system is classified 16 heartbeat types. The measures of proposed system are 85.59%, 95.46%, and 99.56%, for average sensitivity, average selectivity, and average specificity, respectively.

Complexity analysis of the cerebrospinal fluid pulse waveform during infusion studies

PURPOSE

Nonlinear dynamics has enhanced the diagnostic abilities of some physiological signals. Recent studies have shown that the complexity of the intracranial pressure waveform decreases during periods of intracranial hypertension in paediatric patients with acute brain injury. We wanted to assess changes in the complexity of the cerebrospinal fluid (CSF) pressure signal over the large range covered during the study of CSF circulation with infusion studies.

METHODS

We performed 37 infusion studies in patients with hydrocephalus of various types and origin (median age 71 years; interquartile range 60-77 years). After 5 min of baseline measurement, infusion was started at a rate of 1.5 ml/min until a plateau was reached. Once the infusion finished, CSF pressure was recorded until it returned to baseline. We analysed CSF pressure signals using the Lempel-Ziv (LZ) complexity measure. To characterise more accurately the behaviour of LZ complexity, the study was segmented into four periods: basal, early infusion, plateau and recovery.

RESULTS

The LZ complexity of the CSF pressure decreased in the plateau of the infusion study compared to the basal complexity (p=0.0018). This indicates loss of complexity of the CSF pulse waveform with intracranial hypertension. We also noted that the level of complexity begins to increase when the infusion is interrupted and CSF pressure drops towards the initial values.

CONCLUSIONS

The LZ complexity decreases when CSF pressure reaches the range of intracranial hypertension during infusion studies. This finding provides further evidence of a phenomenon of decomplexification in the pulsatile component of the pressure signal during intracranial hypertension.

Sequential algorithm for life threatening cardiac pathologies detection based on mean signal strength and EMD functions

Background

Ventricular tachycardia (VT) and ventricular fibrillation (VF) are the most serious cardiac arrhythmias that require quick and accurate detection to save lives. Automated external defibrillators (AEDs) have been developed to recognize these severe cardiac arrhythmias using complex algorithms inside it and determine if an electric shock should in fact be delivered to reset the cardiac rhythm and restore spontaneous circulation. Improving AED safety and efficacy by devising new algorithms which can more accurately distinguish shockable from non-shockable rhythms is a requirement of the present-day because of their uses in public places.

Method

In this paper, we propose a sequential detection algorithm to separate these severe cardiac pathologies from other arrhythmias based on the mean absolute value of the signal, certain low-order intrinsic mode functions (IMFs) of the Empirical Mode Decomposition (EMD) analysis of the signal and a heart rate determination technique. First, we propose a direct waveform quantification based approach to separate VT plus VF from other arrhythmias. The quantification of the electrocardiographic waveforms is made by calculating the mean absolute value of the signal, called the mean signal strength. Then we use the IMFs, which have higher degree of similarity with the VF in comparison to VT, to separate VF from VTVF signals. At the last stage, a simple rate determination technique is used to calculate the heart rate of VT signals and the amplitude of the VF signals is measured to separate the coarse VF from VF. After these three stages of sequential detection procedure, we recognize the two components of shockable rhythms separately.

Results

The efficacy of the proposed algorithm has been verified and compared with other existing algorithms, e.g., HILB [1], PSR [2], SPEC [3], TCI [4], Count [5], using the MIT-BIH Arrhythmia Database, Creighton University Ventricular Tachyarrhythmia Database and MIT-BIH Malignant Ventricular Arrhythmia Database. Four quality parameters (e.g., sensitivity, specificity, positive predictivity, and accuracy) were calculated to ascertain the quality of the proposed and other comparing algorithms. Comparative results have been presented on the identification of VTVF, VF and shockable rhythms (VF + VT above 180 bpm).

Conclusions

The results show significantly improved performance of the proposed EMD-based novel method as compared to other reported techniques in detecting the life threatening cardiac arrhythmias from a set of large databases.

Classification of arrhythmia using hybrid networks

Reliable detection of arrhythmias based on digital processing of Electrocardiogram (ECG) signals is vital in providing suitable and timely treatment to a cardiac patient. Due to corruption of ECG signals with multiple frequency noise and presence of multiple arrhythmic events in a cardiac rhythm, computerized interpretation of abnormal ECG rhythms is a challenging task. This paper focuses a Fuzzy C- Mean (FCM) clustered Probabilistic Neural Network (PNN) and Multi Layered Feed Forward Network (MLFFN) for the discrimination of eight types of ECG beats. Parameters such as fourth order Auto Regressive (AR) coefficients along with Spectral Entropy (SE) are extracted from each ECG beat and feature reduction has been carried out using FCM clustering. The cluster centers form the input of neural network classifiers. The extensive analysis of Massachusetts Institute of Technology- Beth Israel Hospital (MIT-BIH) arrhythmia database shows that FCM clustered PNNs is superior in cardiac arrhythmia classification than FCM clustered MLFFN with an overall accuracy of 99.05%, 97.14%, respectively.

ECG spectral and morphological parameters reviewed and updated to detect adult and paediatric life-threatening arrhythmia

Since the International Liaison Committee on Resuscitation approved the use of automated external defibrillators (AEDs) in children, efforts have been made to adapt AED algorithms designed for adult patients to detect paediatric ventricular arrhythmias accurately. In this study, we assess the performance of two spectral (A(2) and VFleak) and two morphological parameters (TCI and CM) for the detection of lethal ventricular arrhythmias using an American Heart Association (AHA) compliant database that includes adult and paediatric arrhythmias. Our objective was to evaluate how those parameters can be optimally adjusted to discriminate shockable from nonshockable rhythms in adult and paediatric patients. A total of 1473 records were analysed: 751 from 387 paediatric patients (<or=16 years of age) and 722 records from 381 adult patients. The spectral parameters showed no significant differences (p > 0.01) between the adult and paediatric patients for the shockable records; the differences for nonshockable records however were significant. Still, these parameters maintained the discrimination power when paediatric rhythms were included. A single threshold could be adjusted to obtain sensitivities and specificities above the AHA goals for the complete database. The sensitivities for ventricular fibrillation (VF) and ventricular tachycardia (VT) were 91.1% and 96.6% for VFleak, and 90.3% and 99.3% for A(2). The specificities for normal sinus rhythm (NSR) and other nonshockable rhythms were 99.5% and 96.3% for VFleak, and 99.0% and 97.7% for A(2). On the other hand, the morphological parameters showed significant differences between the adult and paediatric patients, particularly for the nonshockable records, because of the faster heart rates of the paediatric rhythms. Their performance clearly degraded with paediatric rhythms. Using a single threshold, the sensitivities and specificities were below the AHA goals, particularly VT sensitivity (60.4% for TCI and 65.8% for CM) and the specificity for other nonshockable rhythms (51.7% for TCI and 34.5% for CM). The specificities, particularly for the adult case, improve when the thresholds are independently adjusted for each adult and paediatric database.

Distinguishing between ventricular tachycardia and ventricular fibrillation from compressed ECG signal in wireless Body Sensor Networks

Since ECG is huge in size sending large volume data over resource constrained wireless networks is power consuming and will reduce the energy of nodes in Body Sensor Networks (BSN). Therefore, compression of ECGs and diagnosis of diseases from compressed ECGs will play key roles in enhancing the life-time of body sensor networks. Moreover, discrimination between ventricular Tachycardia and Ventricular Fibrillation is of crucial importance to save human life. Existing algorithms work only on plain text ECGs to distinguish between the two, and therefore, not suitable in BSN. VT and VF are often similar in patterns and in filtration of noise and improper attribute selection in compressed ECGs will make it even harder to classify them properly. In this paper, a supervised attribute selection algorithm called Correlation Based Feature Selection (CFS) [4] is used to filter the unwanted attributes and select the most relevant attributes. We then use the selected attributes to train and classify VT and VF using Radial Basis Function (RBF) Neural Network and k-nearest neighbour techniques. We experimented with 103 ECG samples taken from MIT-BIH Malignant Ventricular Ectopy Database. Results showed that accuracy can be as high as 93.3% when attribute selection is used and large number of training samples are provided.

Multistate Lempel-Ziv (MLZ) index interpretation as a measure of amplitude and complexity changes

The Lempel-Ziv complexity (LZ) has been widely used to evaluate the randomness of finite sequences. In general, the LZ complexity has been used to determine the complexity grade present in biomedical signals. The LZ complexity is not able to discern between signals with different amplitude variations and similar random components. On the other hand, amplitude parameters, as the root mean square (RMS), are not able to discern between signals with similar power distributions and different random components. In this work, we present a novel method to quantify amplitude and complexity variations in biomedical signals by means of the computation of the LZ coefficient using more than two quantification states, and with thresholds fixed and independent of the dynamic range or standard deviation of the analyzed signal: the Multistate Lempel-Ziv (MLZ) index. Our results indicate that MLZ index with few quantification levels only evaluate the complexity changes of the signal, with high number of levels, the amplitude variations, and with an intermediate number of levels informs about both amplitude and complexity variations. The study performed in diaphragmatic mechanomyographic signals shows that the amplitude variations of this signal are more correlated with the respiratory effort than the complexity variations. Furthermore, it has been observed that the MLZ index with high number of levels practically is not affected by the existence of impulsive, sinusoidal, constant and Gaussian noises compared with the RMS amplitude parameter.

Fuzzy clustered probabilistic and multi layered feed forward neural networks for electrocardiogram arrhythmia classification

The role of electrocardiogram (ECG) as a noninvasive technique for detecting and diagnosing cardiac problems cannot be overemphasized. This paper introduces a fuzzy C-mean (FCM) clustered probabilistic neural network (PNN) for the discrimination of eight types of ECG beats. The performance has been compared with FCM clustered multi layered feed forward network (MLFFN) trained with back propagation algorithm. Important parameters are extracted from each ECG beat and feature reduction has been carried out using FCM clustering. The cluster centers form the input of neural network classifiers. The extensive analysis using the MIT-BIH arrhythmia database has shown an average classification accuracy of 97.54% with FCM clustered MLFFN and 99.58% with FCM clustered PNN. Fuzzy clustering improves the classification speed as well. The result reveals the capability of the FCM clustered PNN in the computer-aided diagnosis of ECG abnormalities.

Complexity analysis of the fetal heart rate variability: early identification of severe intrauterine growth-restricted fetuses

The main goal of this work is to suggest new indices for a correct identification of the intrauterine growth-restricted (IUGR) fetuses on the basis of fetal heart rate (FHR) variability analysis performed in the antepartum period. To this purpose, we analyzed 59 FHR time series recorded in early periods of gestation through a Hewlett Packard 1351A cardiotocograph. Advanced analysis techniques were adopted including the computation of the Lempel Ziv complexity (LZC) index and the multiscale entropy (MSE), that is, the entropy estimation with a multiscale approach. A multiparametric classifier based on k-mean cluster analysis was also performed to separate pathological and normal fetuses. The results show that the proposed LZC and the MSE could be useful to identify the actual IUGRs and to separate them from the physiological fetuses, providing good values of sensitivity and accuracy (Se = 77.8%, Ac = 82.4%).

Nonlinear dynamical analysis of carbachol induced hippocampal oscillations in mice

AIM

Hippocampal neuronal network and synaptic impairment underlie learning and memory deficit in Alzheimer's disease (AD) patients and animal models. In this paper, we analyzed the dynamics and complexity of hippocampal neuronal network synchronization induced by acute exposure to carbachol, a nicotinic and muscarinic receptor co-agonist, using the nonlinear dynamical model based on the Lempel-Ziv estimator. We compared the dynamics of hippocampal oscillations between wild-type (WT) and triple-transgenic (3xTg) mice, as an AD animal model. We also compared these dynamic alterations between different age groups (5 and 10 months). We hypothesize that there is an impairment of complexity of CCh-induced hippocampal oscillations in 3xTg AD mice compared to WT mice, and that this impairment is age-dependent.

METHODS

To test this hypothesis, we used electrophysiological recordings (field potential) in hippocampal slices.

RESULTS

Acute exposure to 100 micromol/L CCh induced field potential oscillations in hippocampal CA1 region, which exhibited three distinct patterns: (1) continuous neural firing, (2) repeated burst neural firing and (3) the mixed (continuous and burst) pattern in both WT and 3xTg AD mice. Based on Lempel-Ziv estimator, pattern (2) was significantly lower than patterns (1) and (3) in 3xTg AD mice compared to WT mice (P<0.001), and also in 10-month old WT mice compared to those in 5-month old WT mice (P<0.01).

CONCLUSION

These results suggest that the burst pattern (theta oscillation) of hippocampal network is selectively impaired in 3xTg AD mouse model, which may reflect a learning and memory deficit in the AD patients.

Space-time ICA versus Ensemble ICA for ictal EEG analysis with component differentiation via Lempel-Ziv complexity

In this proof-of-principle study we analyzed intracranial electroencephalogram recordings in patients with intractable focal epilepsy. We contrast two implementations of Independent Component Analysis (ICA) - Ensemble (or spatial) ICA (E-ICA) and Space-Time ICA (ST-ICA) in separating out the ictal components underlying the measurements. In each case we assess the outputs of the ICA algorithms by means of a non-linear method known as the Lempel-Ziv (LZ) complexity. LZ complexity quantifies the complexity of a time series and is well suited to the analysis of non-stationary biomedical signals of short length. Our results show that for small numbers of intracranial recordings, standard E-ICA results in marginal improvements in the separation as measured by the LZ complexity changes. ST-ICA using just 2 recording channels both near and far from the epileptic focus result in more distinct ictal components--although at this stage there is a subjective element to the separation process for ST-ICA. Our results are promising showing that it is possible to extract meaningful information from just 2 recording electrodes through ST-ICA, even if they are not directly over the seizure focus. This work is being further expanded for seizure onset analysis.

Complexity analysis of arterial pressure during periods of abrupt hemodynamic changes

In this communication, we estimated the Lempel-Ziv complexity (LZC) on over 40 h of arterial blood pressure (ABP) recordings corresponding to 18 mechanically ventilated animal subjects. In this study, all subjects underwent a period of abrupt hemodynamic changes after an induced injury involving severe blood loss leading to hemorrhagic shock, followed by fluid resuscitation using either lactated ringers or 0.9% normal saline. The LZC metric experienced a statistically significant increase (p < 0.01) immediately following the induced injury and a statistically significant reduction following the administration of fluid therapy (p < 0.01). These results indicate that LZC of ABP may be useful as a dynamic metric to assess fluid responsiveness.

Evolving a Bayesian Classifier for ECG-based Age Classification in Medical Applications

OBJECTIVE: To classify patients by age based upon information extracted from their electro-cardiograms (ECGs). To develop and compare the performance of Bayesian classifiers. METHODS AND MATERIAL: We present a methodology for classifying patients according to statistical features extracted from their ECG signals using a genetically evolved Bayesian network classifier. Continuous signal feature variables are converted to a discrete symbolic form by thresholding, to lower the dimensionality of the signal. This simplifies calculation of conditional probability tables for the classifier, and makes the tables smaller. Two methods of network discovery from data were developed and compared: the first using a greedy hill-climb search and the second employed evolutionary computing using a genetic algorithm (GA). RESULTS AND CONCLUSIONS: The evolved Bayesian network performed better (86.25% AUC) than both the one developed using the greedy algorithm (65% AUC) and the naïve Bayesian classifier (84.75% AUC). The methodology for evolving the Bayesian classifier can be used to evolve Bayesian networks in general thereby identifying the dependencies among the variables of interest. Those dependencies are assumed to be non-existent by naïve Bayesian classifiers. Such a classifier can then be used for medical applications for diagnosis and prediction purposes.

A new algorithm developed based on a mixture of spectral and nonlinear techniques for the analysis of heart rate variability

In this paper, an algorithm based on a joint use of spectral and nonlinear techniques for heart rate variability (HRV) analysis is proposed. First, the measured RR data are passed into a trimmed moving average (TMA)-based filtering system to generate a lower frequency (LF) time series and a higher frequency (HF) one that approximately reflect the sympathetic and vagal activities, respectively. Since the Lyapunov exponent can be used to characterize the level of chaos in complex physiological systems, the largest Lyapunov exponents corresponding to the complex sympathetic and vagal systems are then estimated from the LF and HF time series, respectively, using an existing algorithm. Numerical results of a postural maneuver experiment indicate that both characteristic exponents or their combinations might serve as a set of innovative and robust indicators for HRV analysis, even under the contamination of sparse impulses due to aberrant beats in the RR data.

Analysis of intracranial pressure during acute intracranial hypertension using Lempel-Ziv complexity: further evidence

We analyzed intracranial pressure (ICP) signals during periods of acute intracranial hypertension (ICH) using the Lempel-Ziv (LZ) complexity measure. Our results indicate the LZ complexity of ICP decreases during periods of ICH. The mean LZ complexity before ICH was 0.20+/-0.04, while the mean LZ complexity during ICH was 0.16+/-0.03 (p<0.05). The mean decrease of the LZ complexity values during the ICH episodes was 19.5%. Additionally, we present preliminary evidence suggesting that periods of ICH may be detectable from non-invasive signals coupled with ICP, such as pulse oximetry (SpO2).

Detecting ventricular fibrillation by time-delay methods

A pivotal component in automated external defibrillators (AEDs) is the detection of ventricular fibrillation (VF) by means of appropriate detection algorithms. In scientific literature there exists a wide variety of methods and ideas for handling this task. These algorithms should have a high detection quality, be easily implementable, and work in realtime in an AED. Testing of these algorithms should be done by using a large amount of annotated data under equal conditions. For our investigation we simulated a continuous analysis by selecting the data in steps of 1 s without any preselection. We used the complete BIH-MIT arrhythmia database, the CU database, and files 7001-8210 of the AHA database. For a new VF detection algorithm we calculated the sensitivity, specificity, and the area under its receiver operating characteristic curve and compared these values with the results from an earlier investigation of several VF detection algorithms. This new algorithm is based on time-delay methods and outperforms all other investigated algorithms.

Classification of Heartbeats based on Linear Discriminant Analysis and Artificial Neural Network

In this paper, we proposed a heartbeat classification algorithm based on linear discriminant analysis and artificial neural network. For the input of classifier, we extracted 275 input features from the first derivative signal of ECG signal and RR interval information and it was reduced to be 6 by LDA. To evaluate the performance of the proposed algorithm, we compared the result of the proposed algorithm with that of fuzzy inference system classifier. MIT-BIH Arrhythmia database were used as test and learning data. The performance of the proposed algorithm was 97.49% for sensitivity, 97.91% for specificity and 96.36% for accuracy. For the extraction of features, the first derivative signal of ECG is used only so that the real-time implementation of this algorithm was possible. And, on account of the reduction of feature dimensionality, the time cost for learning and testing can be expected.