T-wave alternans detection by a combined method of principal component analysis and T-wave amplitude.

The detection of T-wave alternans (TWA) has been studied, considering various algorithms based on parameters in the temporal domain. The PhysioNet/Computers in Cardiology Challenge 2008 ('Detecting and quantifying T-wave alternans') provided a dataset consisting of 100 electrocardiographic (ECG) records for the detection or estimation of T-wave alternans. The dataset consists of 68 records from real patients (56 subjects with known risk factors for sudden cardiac death and 12 healthy subjects) and 32 synthetic ECG with calibrated amount of artificial TWA. The participation to this Challenge allowed us to test different algorithms based on two kinds of parameters: (1) the T-wave amplitude computed in a combined lead, and (2) the principal component analysis for quantifying the complexity index of the T-waves. The measurements were made for every beat, and the presence of significant differences between odd and even samples (alternans) was statistically analyzed. Two global indices (considering the beats of the entire record), two local indices (considering sets of sub-intervals) and a proper combination of them are the five indices used for the analysis of TWA detection. Our study proved a high correlation between alternans of T-amplitude and the TWA complexity index. The Challenge scores of our proposed algorithms are all higher than 0.80, and the combined method obtained the highest score among the official participants: 0.890. The PhysioNet/Challenge database was a pilot tentative to build a reference database, which allowed us to make a valuable test of our algorithms.

Assessment and comparison of different methods for heartbeat classification.

The most common way to diagnose cardiac dysfunctions is the ECG signal analysis, usually starting with the assessment of the QRS complex as the most significant wave in the electrocardiogram. Many methods for automatic heartbeats classification have been applied and reported in the literature but the use of different ECG features and the training and testing on different datasets, makes their direct comparison questionable. This paper presents a comparative study of the learning capacity and the classification abilities of four classification methods--Kth nearest neighbour rule, neural networks, discriminant analysis and fuzzy logic. They were applied on 26 morphological parameters, which include information of amplitude, area, interval durations and the QRS vector in a VCG plane and were tested for five types of ventricular complexes--normal heart beats, premature ventricular contractions, left and right bundled branch blocks, and paced beats. One global, one basic and two local learning sets were used. A small-sized learning set, containing the five types of QRS complexes collected from all patients in the MIT-BIH database, was used either with or without applying the leave one out rule, thus representing the global and the basic learning set, respectively. The local learning sets consisted of heartbeats only from the tested patient, which were taken either consecutively or randomly. Using the local learning sets the assessed methods achieved high accuracies, while the small size of the basic learning set was balanced by reduced classification ability. Expectedly, the worst results were obtained with the global learning set.

Premature ventricular contraction classification by the Kth nearest-neighbours rule.

An analysis of electrocardiographic pattern recognition parameters for premature ventricular contraction (PVC) and normal (N) beat classification is presented. Twenty-six parameters were defined: 11 x 2 for the two electrocardiogram (ECG) leads, width of the complex and three parameters derived from a single-plane vectorcardiogram (VCG). Some of the parameters include amplitudes of maximal positive and maximal negative peaks, area of absolute values, area of positive values, area of negative values, number of samples with 70% higher amplitude than that of the highest peak, amplitude and angle of the QRS vector in a VCG plane. They were measured for all heartbeats annotated as N or PVC in all 48 ECG recordings of the MIT-BIH arrhythmia database. Two reference sets for the Kth nearest-neighbours rule were used-global and local. The classification indices obtained with the global reference set were 75.4% specificity and 80.9% sensitivity. Using the local reference set we increased the specificity to 96.7% and the sensitivity to 96.9%. The achieved specificity and sensitivity are comparable with, and greater than, the results reported in the literature.

Ranking of pattern recognition parameters for premature ventricular contractions classification by neural networks.

Detection and classification of ventricular complexes from a limited number of ECG leads is of considerable importance in critical care or operating room patient monitoring. Beat-to-beat detection allows the heart rhythm evolution to be followed and various types of arrhythmia to be recognized. A quantitative analysis is proposed of pattern recognition parameters for classification of normal QRS complexes and premature ventricular contractions (PVC). Twenty-six parameters have been defined: the width of the QRS complex, three vectorcardiogram parameters and 11 from two ECG leads. These parameters include: amplitudes of positive and negative peaks, area of positive and negative waves, various time-interval durations, amplitude and angle of the QRS vector, etc. They are measured for all QRS complexes annotated as 'normals' and 'PVCs' from the 48 ECG recordings of the MIT-BIH arrhythmia database. Neural networks (NN) are shown to be a useful instrument for the analysis of large quantities of parameters. Separate ranking of any parameter and homogeneous group ranking (amplitude, area, interval, slope and vector) were performed. From the two ECG leads, the first three ranked parameter groups for clustering of PVCs are amplitude, slope and interval, while for N clustering they are vector, amplitude and area. Considering the entire parameter set, we obtained N = 99.7% correct detection of normal QRS complexes and PVC = 98.5% of premature ventricular complexes. The study also shows that simultaneous analysis of two ECG channels yields better accuracy compared to using a single channel: the improvement is 0.1% in the classification of N beats and 4.5% for PVC beats.

Automatic detection of atrial fibrillation and flutter by wave rectification method.

Rapid detection of atrial flutter or fibrillation is needed in intensive care or home ECG monitoring with alarm generation, and in portable monitors with warningfunction, etc. Detection and assessment of these atrial abnormalities is necessary in computerized morphological analysis as well, to decide whether parameter measurements should be rejected, restricted to QRS and/or T wave only, or limited to those leads where atrial flutter orfibrillation waves are less expressed. A method for the detection and measurement of atrial flutter and fibrillation in the T-P segments of the ECG is proposed. An atrial flutter/ fibrillation parameter (AFF) is defined as the mean value of the differentiated filtered and rectified signal in these segments. The AFF has been measured in 329 patients from an annotated atrial flutter-fibrillation database. A threshold of AFF=0.35% with respect to the maximum signal excursion was chosen by a heuristic algorithm, to separate patients with atrial arrhythmia. The accuracy of the method was 91.8 %. The positive and negative detection errors of the AFF classification are discussed.

Automatic estimation of the correlation dimension for the analysis of electrocardiograms.

The main purpose of the present work is the definition of a fully automatic procedure for correlation dimension (D(2)) estimation. In the first part, the procedure for the estimation of the correlation dimension (D(2)) is proposed and tested on various types of mathematical models: chaotic (Lorenz and Henon models), periodical (sinusoidal waves) and stochastic (Gaussian and uniform noise). In all cases, accurate D(2) estimates were obtained. The procedure can detect the presence of multiple scaling regions in the correlation integral function. The connection between the presence of multiple scaling regions and multiple dynamic activities cooperating in a system is investigated through the study of composite time series. In the second part of the paper, the proposed algorithm is applied to the study of cardiac electrical activity through the analysis of electrocardiographic signals (ECG) obtained from the commercially available MIT-BIH ECG arrhythmia database. Three groups of ECG signals have been considered: the ECGs of normal subjects and ECGs of subjects with atrial fibrillation and with premature ventricular contraction. D(2) estimates are computed on single ECG intervals (static analysis) of appropriate duration, striking a balance between stationarity requisites and accurate computation requirements. In addition, D(2) temporal variability is studied by analyzing consecutive intervals of ECG tracings (dynamic analysis). The procedure reveals the presence of multiple scaling regions in many ECG signals, and the D(2) temporal variability differs in the three ECG groups considered; it is greater in the case of atrial fibrillation than in normal sinus rhythms. This study points out the importance of considering both the static and dynamic D(2) analysis for a more complete study of the system under analysis. While the static analysis visualizes the underlying heart activity, dynamic D(2) analysis insights the time evolution of the underlying system.

An inference system based on fuzzy logic.

We present the use of fuzzy set theory for the management of imprecision and uncertainty. We first introduced fuzzy set theory according to two different perspectives: the logical and the possibilistic/probabilistic point of view. In addition, several examples of fuzzy sets in different contexts have been considered. The nature of imprecision in the measurement process has been investigated at various levels in order to identify different sources of uncertainty. The fuzzy inference system presented has proved to be a good tool for treating linguistic terms in a quantitative way. An application of a fuzzy inference system in computerized electrocardiography will be described. The main purpose of the present study is to show the potential use of fuzzy logic for the treatment of imprecision and uncertainty.

[Normal electrocardiogram in the aged (the ILSA (Italian Longitudinal Study of Aging) Project)]. / L'elettrocardiogramma normale negli anziani (Studio ILSA).

The ILSA project (Italian Longitudinal Study of Aging) involved acquiring and storing ECG signals and basic clinical-instrumental data via computerized techniques. This database represents the first opportunity to analyze the ECGs of the elderly in Italy and compared a significant sample of the entire Italian population between the ages of 65 and 84. Thus, the study included 88 male subjects and 88 female subjects from four age groups (65-69, 70-74, 75-79, 80-84), for a total of 5632 people. The characteristics of a group of "healthy" patients without any cardiovascular disease were defined using clinical ECG-independent data. After excluding records characterized by an altered morphology (QRS of over 120 ms, confirmed necrotic Q waves in at least two specific leads, confirmed ventricular hypertrophy), a group of 698 "healthy" patients with "normal" ECGs was selected. The mean and standard deviation of the traditional electrocardiographic parameters were evaluated for this group. In agreement with the results of similar studies performed on non-Italian populations, the following conclusions can be drawn: P-wave axis and duration show similar values with respect to the "normal" limits of the adult population; the QRS duration does not show significant changes with age, whereas the QRS axis shifts leftward as age increases; as compared with traditional ranges, the PQ interval is slightly higher than values computed at the corresponding frequency; QT-interval has similar values in the same heart rate interval. The debate on the clinical relevance and meaning of the presence of undoubtedly pathological ECGs in the elderly without any declared or evident pathology still open remains to be settled.

Possibilities of using neural networks for ECG classification.

Some characteristics of the neural network approach have been tested and validated for the particular problem of diagnostic classification in the field of computerized electrocardiography. Two different databases have been used for the evaluation process: CORDA, developed by the Medical Informatics Department of the University of Leuven, and ECG-UCL, developed by the Cliniques Universitaires Saint-Luc, Université Catholique de Louvain. Electrocardiographic signals classified on the basis of electrocardiographic independent clinical data, with a single diagnosis and no conduction abnormalities, have been considered. Seven diagnostic classes have been taken into account, including the different locations of ventricular hypertrophy and myocardial infarction. Two architectures of neural networks have been analyzed in detail considering three aspects: the normalization process, pruning techniques, and fuzzy preprocessing by the use of radial basis functions. The comparison of the results obtained with the two databases will be discussed in detail.

Diagnostic ECG classification based on neural networks.

This study illustrates the use of the neural network approach in the problem of diagnostic classification of resting 12-lead electrocardiograms. A large electrocardiographic library (the CORDA database established at the University of Leuven, Belgium) has been utilized in this study, whose classification is validated by electrocardiographic-independent clinical data. In particular, a subset of 3,253 electrocardiographic signals with single diseases has been selected. Seven diagnostic classes have been considered: normal, left, right, and biventricular hypertrophy, and anterior, inferior, and combined myocardial infarction. The basic architecture used is a feed-forward neural network and the backpropagation algorithm for the training phase. Sensitivity, specificity, total accuracy, and partial accuracy are the indices used for testing and comparing the results with classical methodologies. In order to validate this approach, the accuracy of two statistical models (linear discriminant analysis and logistic discriminant analysis) tuned on the same dataset have been taken as the reference point. Several nets have been trained, either adjusting some components of the architecture of the networks, considering subsets and clusters of the original learning set, or combining different neural networks. The results have confirmed the potentiality and good performance of the connectionist approach when compared with classical methodologies.

Methodology of ECG interpretation in the Padova program.

The main lines of the program designed for the interpretation of ECGs, developed in Padova by LADSEB-CNR with the cooperation of the Medical School of the University of Padova are described. In particular, the strategies used for (i) morphology recognition, (ii) measurement evaluation, and (iii) linguistic decision making are illustrated. The main aspect which discerns this program in comparison with other approaches to computerized electrocardiography is its ability of managing the imprecision in both the measurements and the medical knowledge through the use of fuzzy-set methodologies. So-called possibility distributions are used to represent ill-defined parameters as well as threshold limits for diagnostic criteria. In this way, smooth conclusions are derived when the evidence does not support a crisp decision. The influence of the CSE project on the evolution of the Padova program is illustrated.

A bimolecular mechanism for the cell size control of the cell cycle.

A molecular model for the control of cell size has been developed. It is based on two molecules, one (I) acts as an inhibitor of the entrance into S phase, and it is synthetised just after cell separation in a fixed amount per nucleus. The other (A) is an activator of the S phase, and it is synthetised at a ratio proportional to the overall protein accumulation. The activator reacts stoichiometrically with (I), and after all the (I) molecules have been titrated, (A) begins to accumulate. When it reaches a threshold value, it triggers the onset of DNA replication. This model was tested by simulation and when applied to the case of unequal division explains a number of features of an exponentially growing yeast cell population: (a) the lengths of TP (cycle time of parent cells) and TD (cycle time of daughter cells) verify the condition exp(- KTP ) + exp(- KTD ) = 1; (b) the changes of the average cell size of populations at different growth rates; (c) the frequency of parents and daughters at various growth rates; (d) the increase of cell size at bud initiation for cells of increasing genealogical age; (e) the existence of a TP - TB period (difference between the cycle time of parents and the length of budded phase) that depends linearly upon the doubling time of the population.

Home treatment of seizures as a strategy for the long-term management of febrile convulsions in children.

A cooperative study evaluating compliance and efficacy of Diazepam clisma in home treatment of febrile seizures in children was conducted from January 1979 to June 1981. Parents of 601 children admitted to hospital for a febrile convulsion were taught to use Diazepam clisma in the eventuality of a new seizure and asked to record the length of the episode. Complete follow-up was possible in 564 cases for an average time of 16.7 months. During the research period 109 convulsive episodes were registered in 76 children. Four of these children presented a seizure without fever. Diazepam clisma was administered correctly in 70 episodes (64.2%). In 26 of the remaining 39 cases, therapy was not administered because the seizure ended before the treatment was started. Prolonged seizures (greater than 15 min) have been reported in 8 cases. Six were in the non-treated group and 2 in the treated group. In both these last cases Diazepam was expelled immediately after being administered. The results of the study suggest that home treatment of febrile convulsions by Diazepam clisma represents a well accepted and useful strategy for prevention of prolonged seizures, provided that continuous contact and complete understanding between family and physician can be ensured.