Deep Compressive Sensing on ECG Signals with Modified Inception Block and LSTM

Hua, Jing; Rao, Jue; Peng, Yingqiong; Liu, Jizhong; Tang, Jianjun

doi:10.3390/e24081024

Cited by 14 publications

(5 citation statements)

References 41 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…We further analyze the reconstruction results of the proposed CSML-Net by comparing other popular models (Sun and Feng 2017, Zhang et al 2021a, Hua et al 2022. These models which are specifically designed for CSbased ECG reconstructions are trained using the same hyperparameters.…”

Section: Reconstruction Resultsmentioning

confidence: 99%

CS-based multi-task learning network for arrhythmia reconstruction and classification using ECG signals

Tang

Deng

2023

Physiol. Meas.

View full text Add to dashboard Cite

Although deep learning-based current methods have achieved impressive results in electrocardiograph (ECG) arrhythmia classification issues, they rely on using the original data to identify arrhythmia categories. However, a large amount of data generated by long-term ECG monitoring pose a significant challenge to the limited-bandwidth and real-time systems, which limits the application of deep learning in ECG monitoring. This paper, therefore, proposed a novel multi-task network that combined compressed sensing (CS) and convolutional neural networks, namely CSML-Net. According to the proposed model, the ECG signals were compressed by utilizing a learning measurement matrix and then recovered and classified simultaneously via shared layers and two task branches. Among them, the multi-scale feature module was designed to improve model performance. Experimental results on the MIT-BIH arrhythmia dataset demonstrate that our proposed method is superior to all the approaches that have been compared in terms of reconstruction quality and classification performance. Consequently, the proposed model achieving the reconstruction and classification in the compressed domain can be an improvement and become a promising approach for ECG arrhythmia reconstruction and classification.

show abstract

Section: Reconstruction Resultsmentioning

confidence: 99%

CS-based multi-task learning network for arrhythmia reconstruction and classification using ECG signals

Tang

Deng

2023

Physiol. Meas.

View full text Add to dashboard Cite

show abstract

“…The application of CS on ECG signals is also very effective. Due to the large amount of ECG signal data, the ECG signal can also be reconstructed after using CS compression, which speeds up the data transmission time while ensuring the basic characteristics of the ECG signal and does not affect the medical staff’s judgment of the pathology [ 40 ]. With the continuous development of deep learning, Bora et al have combined CS with generators in neural networks to eliminate the limitation of signal sparsity and speed up the signal reconstruction to some extent, but the overall reconstruction still takes a relatively long time [ 41 ].…”

Section: Literature Reviewmentioning

confidence: 99%

ECG signal classification in wearable devices based on compressed domain

et al. 2023

Self Cite

View full text Add to dashboard Cite

Wearable devices are often used to diagnose arrhythmia, but the electrocardiogram (ECG) monitoring process generates a large amount of data, which will affect the detection speed and accuracy. In order to solve this problem, many studies have applied deep compressed sensing (DCS) technology to ECG monitoring, which can under-sampling and reconstruct ECG signals, greatly optimizing the diagnosis process, but the reconstruction process is complex and expensive. In this paper, we propose an improved classification scheme for deep compressed sensing models. The framework is comprised of four modules: pre-processing; compression; and classification. Firstly, the normalized ECG signals are compressed adaptively in the three convolutional layers, and then the compressed data is directly put into the classification network to obtain the results of four kinds of ECG signals. We conducted our experiments on the MIT-BIH Arrhythmia Database and Ali Cloud Tianchi ECG signal Database to validate the robustness of our model, adopting Accuracy, Precision, Sensitivity and F1-score as the evaluation metrics. When the compression ratio (CR) is 0.2, our model has 98.16% accuracy, 98.28% average accuracy, 98.09% Sensitivity and 98.06% F1-score, all of which are better than other models.

show abstract

“…The ECG data utilized in this study originate from the MIT-BIH Atrial Fibrillation Database (AFDB), detailed in [6,15,16] and the MIT-BIH Arrhythmia Database (MITDB) discussed in research [17][18][19] . The MITDB database serves as a reference for arrhythmias, while the AFDB database is specifically a reference for Atrial Fibrillation (AF).…”

Section: Data Sourcesmentioning

confidence: 99%

Deep Learning Autoencoder Study on ECG Signals

Mochamad Reza,

Satria Mandala,

Zaki

et al. 2023

JNTE

View full text Add to dashboard Cite

Arrhythmia refers to an irregular heart rhythm resulting from disruptions in the heart's electrical activity. To identify arrhythmias, an electrocardiogram (ECG) is commonly employed, as it can record the heart's electrical signals. However, ECGs may encounter interference from sources like electromagnetic waves and electrode motion. Several researchers have investigated the denoising of electrocardiogram signals for arrhythmia detection using deep autoencoder models. Unfortunately, these studies have yielded suboptimal results, indicated by low Signal-to-Noise Ratio (SNR) values and relatively large Root Mean Square Error (RMSE). This study addresses these limitations by proposing the utilization of a Deep LSTM Autoencoder to effectively denoise ECG signals for arrhythmia detection. The model's denoising performance is evaluated based on achieved SNR and RMSE values. The results of the denoising evaluations using the Deep LSTM Autoencoder on the AFDB dataset show SNR and RMSE values of 56.16 and 0.00037, respectively. Meanwhile, for the MITDB dataset, the corresponding values are 65.22 and 0.00018. These findings demonstrate significant improvement compared to previous research. However, it's important to note a limitation in this study—the restricted availability of arrhythmia datasets from MITDB and AFDB. Future researchers are encouraged to explore and acquire a more extensive collection of arrhythmia data to further enhance denoising performance.

show abstract

Deep Compressive Sensing on ECG Signals with Modified Inception Block and LSTM

Cited by 14 publications

References 41 publications

CS-based multi-task learning network for arrhythmia reconstruction and classification using ECG signals

CS-based multi-task learning network for arrhythmia reconstruction and classification using ECG signals

ECG signal classification in wearable devices based on compressed domain

Deep Learning Autoencoder Study on ECG Signals

Contact Info

Product

Resources

About