ECG Segmentation by Neural Networks: Errors and Correction

Sereda, Iana; Alekseev, Sergey; Koneva, Aleksandra; Kataev, Roman; Osipov, Grigory V.

doi:10.1109/ijcnn.2019.8852106

Cited by 25 publications

(24 citation statements)

References 12 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…In a deep hierarchical structure, the learned features tend to become more abstract as the network gets deeper [38]. Convolutional neural networks (CNNs), including hidden layers of convolutional filters with pretrained weights can be run in real-time, thus being feasible for different ECG monitoring applications, such as denoising [39,40], QRS detection [41,42], ECG segmentation [43], heartbeat classification [44][45][46][47][48], and arrhythmia classification with different output diagnosis labels (normal rhythm, atrial fibrillation, other rhythm, noise [49][50][51][52]; normal rhythm, atrial fibrillation, atrial flutter, ventricular fibrillation [53,54]). While the above studies use DNN architectures with 3 to 11 hidden layers, a recent study of Hannun et al (2019) [55] has demonstrated that an end-to-end 34-layer DNN can classify a broad range of 12 distinct arrhythmias with high diagnostic performance similar to that of cardiologists.…”

Section: Introductionmentioning

confidence: 99%

Fully Convolutional Deep Neural Networks with Optimized Hyperparameters for Detection of Shockable and Non-Shockable Rhythms

Krasteva

Ménétré

Didon

et al. 2020

Sensors

View full text Add to dashboard Cite

Deep neural networks (DNN) are state-of-the-art machine learning algorithms that can be learned to self-extract significant features of the electrocardiogram (ECG) and can generally provide high-output diagnostic accuracy if subjected to robust training and optimization on large datasets at high computational cost. So far, limited research and optimization of DNNs in shock advisory systems is found on large ECG arrhythmia databases from out-of-hospital cardiac arrests (OHCA). The objective of this study is to optimize the hyperparameters (HPs) of deep convolutional neural networks (CNN) for detection of shockable (Sh) and nonshockable (NSh) rhythms, and to validate the best HP settings for short and long analysis durations (2–10 s). Large numbers of (Sh + NSh) ECG samples were used for training (720 + 3170) and validation (739 + 5921) from Holters and defibrillators in OHCA. An end-to-end deep CNN architecture was implemented with one-lead raw ECG input layer (5 s, 125 Hz, 2.5 uV/LSB), configurable number of 5 to 23 hidden layers and output layer with diagnostic probability p ∈ [0: Sh,1: NSh]. The hidden layers contain N convolutional blocks × 3 layers (Conv1D (filters = Fi, kernel size = Ki), max-pooling (pool size = 2), dropout (rate = 0.3)), one global max-pooling and one dense layer. Random search optimization of HPs = {N, Fi, Ki}, i = 1, … N in a large grid of N = [1, 2, … 7], Fi = [5;50], Ki = [5;100] was performed. During training, the model with maximal balanced accuracy BAC = (Sensitivity + Specificity)/2 over 400 epochs was stored. The optimization principle is based on finding the common HPs space of a few top-ranked models and prediction of a robust HP setting by their median value. The optimal models for 1–7 CNN layers were trained with different learning rates LR = [10−5; 10−2] and the best model was finally validated on 2–10 s analysis durations. A number of 4216 random search models were trained. The optimal models with more than three convolutional layers did not exhibit substantial differences in performance BAC = (99.31–99.5%). Among them, the best model was found with {N = 5, Fi = {20, 15, 15, 10, 5}, Ki = {10, 10, 10, 10, 10}, 7521 trainable parameters} with maximal validation performance for 5-s analysis (BAC = 99.5%, Se = 99.6%, Sp = 99.4%) and tolerable drop in performance (<2% points) for very short 2-s analysis (BAC = 98.2%, Se = 97.6%, Sp = 98.7%). DNN application in future-generation shock advisory systems can improve the detection performance of Sh and NSh rhythms and can considerably shorten the analysis duration complying with resuscitation guidelines for minimal hands-off pauses.

show abstract

Section: Introductionmentioning

confidence: 99%

Fully Convolutional Deep Neural Networks with Optimized Hyperparameters for Detection of Shockable and Non-Shockable Rhythms

Krasteva

Ménétré

Didon

et al. 2020

Sensors

View full text Add to dashboard Cite

show abstract

“…We developed and tested an algorithm for segmentation and classification of 12-lead ECGs. Even though our approach for segmentation did not achieve a state of the art result in beat detection, its performance is comparable to other deep learning based approaches [23][24][25].…”

Section: Discussionmentioning

confidence: 63%

ECG Segmentation using a Neural Network as the Basis for Detection of Cardiac Pathologies

Sodmann¹,

Vollmer²

2020

2020 Computing in Cardiology Conference (CinC)

View full text Add to dashboard Cite

Electrocardiography allows fast and noninvasive diagnosis and screening of a wide range of cardiac diseases. The interpretation of ECGs is difficult and depends on the levels of training of the physician. In consequence, pathologies can remain undiagnosed or norm-variations are interpreted as pathological. The PhysioNet/Computing in Cardiology Challenge 2020 aims to classify various cardiac pathologies in 12lead ECGs, data was collected across a variety of different clinics and countries to pave the way for a common evaluation of ECGs. Our Team Heartly-AI proposes a two-step algorithm using a UNet and XGBoost for the 2020 Phys-ioNet Computing in Cardiology Challenge "Classification of 12 lead ECGs". The algorithm achieved a 5-fold crossvalidation metric of 0.113 and scored 0.136 on the official test set, therefore placing us 28th out of the 41 teams in the official ranking.

show abstract

“…networks of the "linear" multilayer convolutional structure without complications like in [18]. But it was shown that even such basic architectures show a satisfactory result on the same ECG segmentation task [19], and very good results -in ECG-based biometric authentication systems [20]. Segmentation results on healthy patients even in such networks were high, although the result was somewhat worse on unhealthy ones.…”

Section: Discussionmentioning

confidence: 99%

Problems of representation of electrocardiograms in convolutional neural networks

Sereda

Alekseev

Koneva

et al. 2020

2020 International Joint Conference on Neural Networks (IJCNN)

Self Cite

View full text Add to dashboard Cite

Using electrocardiograms as an example, we demonstrate the characteristic problems that arise when modeling one-dimensional signals containing inaccurate repeating pattern by means of standard convolutional networks. We show that these problems are systemic in nature. They are due to how convolutional networks work with composite objects, parts of which are not fixed rigidly, but have significant mobility. We also demonstrate some counterintuitive effects related to generalization in deep networks.

show abstract

ECG Segmentation by Neural Networks: Errors and Correction

Cited by 25 publications

References 12 publications

Fully Convolutional Deep Neural Networks with Optimized Hyperparameters for Detection of Shockable and Non-Shockable Rhythms

Fully Convolutional Deep Neural Networks with Optimized Hyperparameters for Detection of Shockable and Non-Shockable Rhythms

ECG Segmentation using a Neural Network as the Basis for Detection of Cardiac Pathologies

Problems of representation of electrocardiograms in convolutional neural networks

Contact Info

Product

Resources

About