Classification of ECG Using Ensemble of Residual CNNs with Attention Mechanism

Nejedlý, Petr; Ivora, Adam; Smíšek, Radovan; Viščor, Ivo; Koscova, Zuzana; Jurák, Pavel; Plešinger, Filip

doi:10.23919/cinc53138.2021.9662723

Cited by 32 publications

(27 citation statements)

References 7 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The unique class that was not modified in comparison to the original challenge's dataset is the AF, enabling some degree of comparison. The challenge winner, ISIBrno-AIMT team, used an ensemble of residual networks approach that achieved 97.1% AUROC and an 83% F1-score when using only the leads I and II [15]. Considering only the AF class (and two lead approaches), the snu adsl team achieved the best AUROC (98.0%) and F1-score (88.3%) by using representational learning and an EfficientNet-B3 model [13].…”

Section: Resultsmentioning

confidence: 99%

“…Residual networks (and variants) were very common in the "Will two do?" challenge [14,15,21] and other ECG classification tasks [22,23]. And as our proposed model is an adaptation of the VGG-11, the original model is used to compare the impact of the reduction in complexity.…”

Section: Training and Evaluationmentioning

confidence: 99%

“…However, the computational complexity of most classification models, particularly those based on deep learning algorithms, is one of the challenges for using wearable devices in continuous rhythm monitoring. To improve accuracy, these models frequently use deeper and more complex models [13,14] or ensembles of models [15], usually without taking into account the energy consumption or computational limitations of portable devices [16].…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

A Lightweight Unidimensional Deep Learning Model for Atrial Fibrillation Detection

Soares¹,

Monteiro²,

Jatene³

et al. 2022

Computing in Cardiology Conference (CinC)

View full text Add to dashboard Cite

Continuous rhythm monitoring using wearable devices is a potential tool for early identification of atrial fibrillation (AF), the most frequent cardiac arrhythmia (with 0,51% worldwide prevalence, increasing with time), and is also a tool for remote monitoring patients after cardiac surgery. However, AF detection directly through wearable devices is limited by the computational complexity of the classifier model.In this work we propose a lightweight AF classifier model based on the VGG-11 architecture (LiteVGG-11), focusing on reducing the number of parameters and numerical operations. Using a low number of filters, depthwise separable convolution, and global pooling, this model has only 20,454 parameters and needs 6.9 MFLOP to make an inference for an input of 10 seconds of the ECG leads I and II, sampled at 200 Hz.To test its effectiveness for AF detection we used the PhysioNet/CinC Challenge 2021 public dataset, stratifying the classes into sinus rhythm, AF, and other rhythms. After 10 Monte Carlo cross-validation splits, with 24,260 unbalanced samples for training and 1,536 balanced samples for validation and testing, the observed metrics (mean±standard deviation) were: Se 94.1±0.1%; Sp 91.9±0.8%; F1-Score 89.50.7±%; and AUC 96.1±0.6%.

show abstract

Section: Resultsmentioning

confidence: 99%

Section: Training and Evaluationmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

A Lightweight Unidimensional Deep Learning Model for Atrial Fibrillation Detection

Soares¹,

Monteiro²,

Jatene³

et al. 2022

Computing in Cardiology Conference (CinC)

View full text Add to dashboard Cite

show abstract

“…Probability scores are obtained by applying P 1 (sigmoid) to the output from logits block. Then, a differential evolution genetic technique is used to optimize class thresholds [18]. Our experiments reveal that these thresholds do not vary as the number of leads are varied.…”

Section: Parallel Mlkd (P-mlkd)mentioning

confidence: 94%

LRH-Net: A Multi-Level Knowledge Distillation Approach for Low-Resource Heart Network

Chauhan¹,

Guptha²,

Reddy³

et al. 2022

Preprint

View full text Add to dashboard Cite

An electrocardiogram (ECG) monitors the electrical activity generated by the heart and is used to detect fatal cardiovascular diseases (CVDs). Conventionally, to capture the precise electrical activity, clinical experts use multiple-lead ECGs (typically 12 leads). But in recent times, large-size deep learning models have been used to detect these diseases. However, such models require heavy compute resources like huge memory and long inference time. To alleviate these shortcomings, we propose a low-parameter model, named Low Resource Heart-Network (LRH-Net), which uses fewer leads to detect ECG anomalies in a resource-constrained environment. A multi-level knowledge distillation process is used on top of that to get better generalization performance on our proposed model. The multi-level knowledge distillation process distills the knowledge to LRH-Net trained on a reduced number of leads from higher parameter (teacher) models trained on multiple leads to reduce the performance gap. The proposed model is evaluated on the PhysioNet-2020 challenge dataset with constrained input. The parameters of the LRH-Net are 106x less than our teacher model for detecting CVDs. The performance of the LRH-Net was scaled up to 3.2% and the inference time scaled down by 75% compared to the teacher model. In contrast to the compute-and parameter-intensive deep learning techniques, the proposed methodology uses a subset of ECG leads using the low resource LRH-Net, making it eminently suitable for deployment on edge devices.

show abstract

“…The model classifies the standard 12-lead ECG signal into 30 rhythm types and is derived from our winning submission to the CinC/PhysioNet 2021 challenge. 21 However, it does not behave precisely the same way since the original solution consisted of three separately trained models, forming an ensemble. The model uses the ResNet architecture with an attention mechanism.…”

Section: Ecg Classification Model (Ecgclassification12)mentioning

confidence: 99%

DeepPlayer: An open‐source SignalPlant plugin for deep learning inference

et al. 2022

View full text Add to dashboard Cite

Background and Objective: Machine learning has become a powerful tool in several computation domains. The most progressive way of machine learning, deep learning, has already surpassed several algorithms designed by human experts. It also applies to the field of biomedical signal processing. However, while many experts produce deep learning models, there is no software platform for signal processing, allowing the convenient use of pre-trained deep learning models and evaluating them using any inspected signal. This may also hinder understanding, interpretation, and explanation of results. For these reasons, we designed DeepPlayer. It is a plugin for the free signal processing software SignalPlant. The plugin allows loading deep learning models saved in the Open Neural Network Exchange (ONNX) file format and evaluating them on any given signal. Methods:The DeepPlayer plugin and its graphical user interface were designed in C# programming language and the .NET framework. We used the inference library OnnxRuntime, which supports graphics card acceleration. The inference is executed in asynchronous tasks for a live preview and evaluation of the signals. Model outputs can be exported back to SignalPlant for further processing, such as peak detection or thresholding. Results:We developed the DeepPlayer plugin to evaluate deep learning models in SignalPlant.The plugin keeps with SignalPlant's interactive work with signals, such as live preview or easy selection of associated signals. The plugin can load classification or regression models and allows standard pre-processing and post-processing methods. We prepared several deep learning models to test the plugin. Additionally, we provide a tutorial training script that outputs an ONNX format model with correctly set metadata information. These, and the source code of the DeepPlayer plugin, are publicly accessible via GitHub and Google Colab service. Conclusion:The DeepPlayer plugin allows running deep learning models easily and interactively. Therefore, experts and non-AI experts alike can explore and apply deep learning models for (biomedical) signal processing. Its ease of use and interactivity might also contribute to a better understanding and acceptance of AI methods in biomedicine.

show abstract

Classification of ECG Using Ensemble of Residual CNNs with Attention Mechanism

Cited by 32 publications

References 7 publications

A Lightweight Unidimensional Deep Learning Model for Atrial Fibrillation Detection

A Lightweight Unidimensional Deep Learning Model for Atrial Fibrillation Detection

LRH-Net: A Multi-Level Knowledge Distillation Approach for Low-Resource Heart Network

DeepPlayer: An open‐source SignalPlant plugin for deep learning inference

Contact Info

Product

Resources

About