Physikalisch-Technische Bundesanstalt Diagnostic Database Research Articles

Localization of myocardial infarction using a multi-branch weight sharing network based on 2-D vectorcardiogram

Early diagnosis and localization of myocardial infarction (MI) assist clinicians in saving numerous lives through the timely treatment for patients with MI. Vectorcardiogram (VCG) can reflect the characteristic changes of cardiac electrical activity in MI in detail. In this context, the present study reports a multi-branch weight sharing network model based on 2-D VCG constructed to realize the automatic localization of MI. The three-branch network extracted the spatial morphological features of the three planes of the 2-D VCG, respectively, and the weight-sharing part of the network obtained the spatial correlation information among the three planes. Subsequently, the Softmax classifier was employed to classify normal individuals and MI patients (11 class infarct sites). To evaluate the performance of the proposed method for MI localization, PTB(Physikalisch-Technische Bundesanstalt) diagnostic ECG database was employed. The localization accuracy, sensitivity, and specificity achieved using the proposed method were 99.87%, 99.92%, and 99.99%, respectively. Thus, the proposed scheme is expected to be useful in assisting cardiologists in interpreting VCG for clinical diagnosis.

An effective PSR-based arrhythmia classifier using self-similarity analysis

Among different cardiac arrhythmias, Ventricular Arrhythmias (VA) are fatal and life-threatening. Therefore, the detection and classification of VA is crucial task for cardiologists. However, in some cases, the ECG morphologies of two kinds of VA - Ventricular Tachycardia (VT) and Ventricular Fibrillation (VF) are similar and difficult to distinguish by human eyes. In this study, we present a low computational complexity arrhythmia classifier with high accuracy based on Phase Space Reconstruction (PSR). It is used to classify normal electrocardiogram (ECG), atrial fibrillation (AF), VT, VF and VT followed by VF. The Creighton University Ventricular Tachyarrhythmia Database (CUDB), Physikalisch-Technische Bundesanstalt Diagnostic Database (PTBDB), MIT-BIH Atrial Fibrillation Database (MIT-BIH AFDB) from PhysioNet databank and University Hospital Southampton database (UHSDB) are used for evaluation and comparison of the proposed algorithm. Two PSR diagrams were plotted based on a window length of 5 s ECG with two different time delays and the PSR-based features were extracted from them using the box-counting technique. This process was applied on 122 records with more than 5500 windows of ECG signals. The results show an average sensitivity of 98.73%, specificity of 99.71% and accuracy of 99.56%. The average computational time of our proposed method for one 5 s window processing is 1.9 s and therefore has the potential in real-time arrhythmia classification applications.

Automated Detection of Posterior Myocardial Infarction From Vectorcardiogram Signals Using Fourier–Bessel Series Expansion Based Empirical Wavelet Transform

Posterior myocardial infarction (PMI) is a fatal condition of the human heart, wherein the posterior coronary circulation becomes disrupted. If left untreated, the PMI can cause a severe heart attack. To detect PMI, the conventional 12-lead standard electrocardiogram signals demonstrate poor detection sensitivity due to the absence of posterior sensing electrodes. Contrastingly, the three lead vectorcardiogram (VCG) signal has the sensor electrodes placed on the posterior side, thereby improving the reliability of PMI diagnosis. In this letter, we use VCG signals for the classification of PMI and healthy control (HC) subjects. The abnormalities in the electrical conduction due to PMI are captured using Fourier–Bessel series expansion based empirical wavelet transform, followed by the singular value decomposition (SVD). The resultant feature space is then utilized to classify PMI and HC segments using a support vector machine (SVM) classifier with various kernels, namely, linear, Gaussian, and radial basis function (RBF). The SVM with RBF kernel has resulted in the maximum classification accuracy, sensitivity, and specificity of 95.52, 91.08, and 97.45%, respectively, over the Physikalisch-Technische Bundesanstalt diagnostic database. Thus, the proposed method has the potential to be used in the clinical setting for accurate and robust PMI detection.

MFB-CBRNN: A Hybrid Network for MI Detection Using 12-Lead ECGs.

This paper proposes a novel hybrid network named multiple-feature-branch convolutional bidirectional recurrent neural network (MFB-CBRNN) for myocardial infarction (MI) detection using 12-lead ECGs. The model efficiently combines convolutional neural network-based and recurrent neural network-based structures. Each feature branch consists of several one-dimensional convolutional and pooling layers, corresponding to a certain lead. All the feature branches are independent from each other, which are utilized to learn the diverse features from different leads. Moreover, a bidirectional long short term memory network is employed to summarize all the feature branches. Its good ability of feature aggregation has been proved by the experiments. Furthermore, the paper develops a novel optimization method, lead random mask (LRM), to alleviate overfitting and implement an implicit ensemble like dropout. The model with LRM can achieve a more accurate MI detection. Class-based and subject-based fivefold cross validations are both carried out using Physikalisch-Technische Bundesanstalt diagnostic database. Totally, there are 148 MI and 52 healthy control subjects involved in the experiments. The MFB-CBRNN achieves an overall accuracy of 99.90% in class-based experiments, and an overall accuracy of 93.08% in subject-based experiments. Compared with other related studies, our algorithm achieves a comparable or even better result on MI detection. Therefore, the MFB-CBRNN has a good generalization capacity and is suitable for MI detection using 12-lead ECGs. It has a potential to assist the real-world MI diagnostics and reduce the burden of cardiologists.

Myocardial infarction detection system from PTB diagnostic ECG database using Fuzzy inference system for S-T waves

Heart disease was the one of major health problem in the world, which caused many deaths. Myocardial infarction (MI) is one type of heart disease which caused by a blockage in the coronary arteries. This disease could be detected by reading electrocardiogram (ECG) wave result. Knowledge and expert analysis were required to read PQRST-wave in Electrocardiograms. Fuzzy inference system was used in this detection system because it flexibility on linguistic variables. Fuzzy inference system could be performed after discovery S and T peaks. The characteristics of Myocardial Infarction could be seen through the condition of S and T wave. Detection system test was conducted on databases that obtained from Physionet bank, Physikalisch-Technische Bundesanstalt (PTB) diagnostic ECG database that were collected from 52 healthy patients and 148 diagnosed MI patients. The result of this research showed that test of detection system had sensitivity level of 73%.

Shannon’s Energy Based Algorithm in ECG Signal Processing

Physikalisch-Technische Bundesanstalt (PTB) database is electrocardiograms (ECGs) set from healthy volunteers and patients with different heart diseases. PTB is provided for research and teaching purposes by National Metrology Institute of Germany. The analysis method of complex QRS in ECG signals for diagnosis of heart disease is extremely important. In this article, a method on Shannon energy (SE) in order to detect QRS complex in 12 leads of ECG signal is provided. At first, this algorithm computes the Shannon energy (SE) and then makes an envelope of Shannon energy (SE) by using the defined threshold. Then, the signal peaks are determined. The efficiency of the algorithm is tested on 70 cases. Of all 12 standard leads, ECG signals include 840 leads of the PTB Diagnostic ECG Database (PTBDB). The algorithm shows that the Shannon energy (SE) sensitivity is equal to 99.924%, the detection error rate (DER) is equal to 0.155%, Positive Predictivity (+P) is equal to 99.922%, and Classification Accuracy (Acc) is equal to 99.846%.