Atrial Fibrillation Detection in Short Single Lead ECG Recordings Using Wavelet Transform and Artificial Neural Networks

Detection and classification of atrial and ventricular cardiovascular diseases to improve the cardiac health literacy for resource constrained regions

ECG is a non-invasive way of determining cardiac health by measuring the electrical activity of the heart. A novel detection technique for feature points P, QRS and T is investigated to diagnose various atrial and ventricular cardiovascular anomalies with ECG signals for ambulatory monitoring. Before the system is worthy of field trials, it is validated with several databases and recorded their response. The QRS complex detection is based on the Pan Tompkins algorithm and difference operation method that provides positive predictivity, sensitivity and false detection rate of 99.29%, 99.49% and 1.29%, respectively. Proposed novel T wave detection provides sensitivity of 97.78%. Also, proposed P wave detection provides positive predictivity, sensitivity and false detection rate of 99.43%, 99.4% and 1.15% for the control study (normal subjects) and 82.68%, 94.3% and 25.4% for the case (patients with cardiac anomalies) study, respectively. Disease detection such as arrhythmia is based on standard R-R intervals while myocardial infarction is based on the ST-T deviations where the positive predictivity, sensitivity and accuracy are observed to be 94.6%, 84.2% and 85%, respectively. It should be noted that, since the frontal leads are only used, the anterior myocardial infarction cases are detected with the injury pattern in lead avl and ST depression in reciprocal leads. Detection of atrial fibrillation is done for both short and long duration signals using statistical methods using interquartile range and standard deviations, giving very high accuracy, 100% in most cases. The system hardware for obtaining the 2 lead ECG signal is designed using commercially available off the shelf components. Small field validation of the designed system is performed at a Public Health Centre in Gujarat, India with 42 patients (both cases and controls). 78.5% accuracy was achieved during the field validation. It is thus concluded that the proposed method is ideal for improvisation in cardiac health monitoring outreach in resource constrained regions.

Multiscale Encoding of Electrocardiogram Signals with a Residual Network for the Detection of Atrial Fibrillation

Atrial fibrillation (AF) is one of the most common cardiac arrhythmias, and it is an indication of high-risk factors for stroke, myocardial ischemia, and other malignant cardiovascular diseases. Most of the existing AF detection methods typically convert one-dimensional time-series electrocardiogram (ECG) signals into two-dimensional representations to train a deep and complex AF detection system, which results in heavy training computation and high implementation costs. In this paper, a multiscale signal encoding scheme is proposed to improve feature representation and detection performance without the need for using any transformation or handcrafted feature engineering techniques. The proposed scheme uses different kernel sizes to produce the encoded signal by using multiple streams that are passed into a one-dimensional sequence of blocks of a residual convolutional neural network (ResNet) to extract representative features from the input ECG signal. This also allows networks to grow in breadth rather than in depth, thus reducing the computing time by using the parallel processing capability of deep learning networks. We investigated the effects of the use of a different number of streams with different kernel sizes on the performance. Experiments were carried out for a performance evaluation using the publicly available PhysioNet CinC Challenge 2017 dataset. The proposed multiscale encoding scheme outperformed existing deep learning-based methods with an average F₁ score of 98.54%, but with a lower network complexity.

Variable step dynamic threshold local binary pattern for classification of atrial fibrillation

OBJECTIVE

In this paper, we proposed new methods for feature extraction in machine learning-based classification of atrial fibrillation from ECG signal.

METHODS

Our proposed methods improved conventional 1-dimensional local binary pattern method in two ways. First, we proposed a dynamic threshold LBP code generation method for use with 1-dimensional signals, enabling the generated LBP codes to have a more detailed representation of the signal morphological pattern. Second, we introduced a variable step value into the LBP code generation algorithm to better cope with a high sampling frequency input signal without a downsampling process. The proposed methods do not employ computationally expensive processes such as filtering, wavelet transform, up/downsampling, or beat detection, and can be implemented using only simple addition, division, and compare operations.

RESULTS

Combining these two approaches, our proposed variable step dynamic threshold local binary pattern method achieved 99.11% sensitivity and 99.29% specificity when used as a feature generation algorithm in support vector machine classification of atrial fibrillation from MIT-BIH Atrial Fibrillation Database dataset. When applied on signals from MIT-BIH Arrhythmia Database, our proposed method achieved similarly good 99.38% sensitivity and 98.97% specificity.

CONCLUSION

Our proposed methods achieved one of the best results among published works in atrial fibrillation classification using the same dataset while using less computationally expensive calculations, without significant performance degradation when applied on signals from multiple databases with different sampling frequencies.