Deep Learning Approach for QRS Wave Detection in ECG Monitoring

G2-ResNeXt: A Novel Model for ECG Signal Classification

“…At present, most of research on heartbeat classification relatively depends on the accuracy of QRS wave detection. As the characteristics of the human body and the form and parameters of a single signal are variable, certain difference may exist in the positioning results of different QRS wave detection algorithms under the same accuracy [13]. For example, some QRS algorithms provide position slightly to the left, while others provide position slightly to the right, resulting in an obvious deviation in the final positioning results even if the same data are processed.…”

“…Because the label is in the form of One-Hot encoding, the value in the label of a certain sample type is 1 in the corresponding position, and the rest are 0. The optimized formula is shown below: 𝐹𝐿 = −𝛼 𝑐 (1 − 𝑝 𝑐 ) 𝛾 log(𝑝 𝑐 ) , (13) where 𝛼 𝑐 denotes the weight of the class-𝑐 sample and 𝑝 𝑐 denotes the probability value of the class-𝑐 output produced by SoftMax.…”

1D neural network design to detect cardiac arrhythmias

“…However, in some cases have been designed multichannel pattern networks (MART) that allow more than one input of the processed signal [9]. Some studies seek to classify rhythms that have an almost defined pattern with symmetry functions to extract patterns [10] or characteristics of the ECG signal such as duration, amplitude, gradients, among others [11], also seeking the classification of sinus arrhythmias or ventricular arrhythmias from individual analysis [12][13], however, other networks seek to establish a significant difference between a specific wave compared to normal waves (sinus rhythms), such as efficiently detecting a blockage or ischemic episodes [14][15] and premature ventricular and atrial onset [16]; The most common rhythm is atrial fibrillation and therefore it is considered important to establish classification and prediction algorithms for it [17] as Artis, Mark and Moody did, using the Markov model for the implementation of a neural network based on the Back-Propagation algorithm and trained with signals obtained from the MIT-BIH database, present in Physionet® [18][19].…”