Multi-Label Attribute Selection of Arrhythmia for Electrocardiogram Signals with Fusion Learning

Abstract

There are three primary challenges in the automatic diagnosis of arrhythmias by electrocardiogram (ECG): the significant variation among individual patients, the multiple pathologies in the ECG signal and the high cost in annotating clinical ECG with the corresponding labels. Traditional ECG processing approaches rely heavily on prior knowledge, such as those from feature extraction and waveform analysis. The preprocessing for prior knowledge incurs computational overhead. Furthermore, standard deep learning methods do not fully consider the dynamic temporal, spatial and multi-labeling characteristics of ECG data. In clinical ECG waveforms, it is common to see multi-labeling in which a patient is labeled with multiple classes of arrhythmias. However, multiclass approaches in current research mainly solve the multi-label machine learning problem, ignoring the correlation between diseases, resulting in information loss. In this paper, an arrhythmia detection and classification scheme called multi-label fusion deep learning is proposed. The objective is to build a unified system with automatic feature learning which supports effective multi-label classification. First, a multi-label ECG-based feature selection method is combined with a matrix decomposition and sparse learning theory. The optimal feature subset is selected as a preprocessing algorithm for ECG data. A multi-label classifier is then constructed by fusing CNN and RNN networks to fully exploit the interactions and features of the time and space dimensions. The experimental result demonstrates that the proposed method can achieve a state-of-the-art performance compared to other algorithms in multi-label database experiments.

ECG signal processing by feature selection with classification based on machine learning architectures in cardiovascular disease detection

“…Assuming that F is a collection of chosen features, c is class label, and fi (1 ≤ i < |F|) is any feature in F. Eq. 8 [12,13] can be used to assess the relevance between F and c, which implies that Eq. 9 can be used to demonstrate the contribution of fi to the relevance.…”

“…12 is therefore suggested here to calculate the quality of fi in F, which is the relevance of fi minus the redundancy it adds. The size of the feature subset F is denoted by |F| represented by eqn (13).…”

“…In their investigation, the authors of one of these obtained outcomes utilizing ResNet method at levels of AUC = 0.967 and ACC = 0.43 [12]. A similar strategy was put forth [13], employing SE_ResNet method to increase effectiveness of categorising different ECG anomalies. Others evaluated the viability of utilising CNNs, particularly those based on ResNet as well as Inception architectures, focusing on comparative analysis of recently available PTB-XL dataset [14].…”

Multi-Label Diagnosis of Arrhythmias Based on a Modified Two-Category Cross-Entropy Loss Function

“…For example, in ECG signals of a period of sustained atrial fibrillation, PVCs but not premature atrial fibrillation can occur simultaneously. The relationship between the various designations is complex, making multi-label classification of ECG signals challenging [22][23][24]. Yoo et al [25] optimized the algorithm from the perspective of multi-label classification of arrhythmia and proposed xECGNet.…”