Temporal feature and heuristics-based Noise Detection over Classical Machine Learning for ECG Signal Quality Assessment

Hermawan, Indra; Ma’sum, M. Anwar; Intan, Purnamasari; Jatmiko, Wisnu; Wiweko, Budi; Boediman, Alfred; Pradekso, Beno K.

doi:10.1109/iwbis.2019.8935757

Cited by 6 publications

(3 citation statements)

References 12 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…As shown in Figure 3 , the CinC11 [ 4 ] dataset was most frequently used in the studies; it originated from the PhysioNet/Computing in Cardiology Challenge 2011. Eleven studies only cross-validated their results on the same dataset that they used to train their model [ 6 , 8 , 10 , 11 , 12 , 14 , 19 , 22 , 24 , 25 , 29 ]. This likely results in a higher accuracy than cross-validating on data gathered separately from the training set.…”

Section: Discussionmentioning

confidence: 99%

Automatic ECG Quality Assessment Techniques: A Systematic Review

2022

View full text Add to dashboard Cite

Cardiovascular diseases are the leading cause of death, globally. Stroke and heart attacks account for more than 80% of cardiovascular disease-related deaths. To prevent patient mismanagement and potentially save lives, effective screening at an early stage is needed. Diagnosis is typically made using an electrocardiogram (ECG) analysis. However, ECG recordings are often corrupted by different types of noise, degrading the quality of the recording and making diagnosis more difficult. This paper reviews research on automatic ECG quality assessment techniques used in studies published from 2012–2022. The CinC11 Dataset is most often used for training and testing algorithms. Only one study tested its algorithm on people in real-time, but it did not specify the demographic data of the subjects. Most of the reviewed papers evaluated the quality of the ECG recordings per single lead. The accuracy of the algorithms reviewed in this paper range from 85.75% to 97.15%. More clarity on the research methods used is needed to improve the quality of automatic ECG quality assessment techniques and implement them in a clinical setting. This paper discusses the possible shortcomings in current research and provides recommendations on how to advance the field of automatic ECG quality assessment.

show abstract

Section: Discussionmentioning

confidence: 99%

Automatic ECG Quality Assessment Techniques: A Systematic Review

2022

View full text Add to dashboard Cite

show abstract

“…On the opposite, it decreases the performance of the model. [7] No RBF-SVM 100 33 57.446 [7] No MLP 100 33 57.446 [7] No R.Forest 100 31 55.678…”

Section: F Performance Comparisonmentioning

confidence: 99%

“…Hermawan et al proposed ECG signal quality assessment using temporal feature and heuristic-based (rule-based) method [6]. Hermawan et al also measure the performance of the method compared with fully supervised classical machine learning method [7]. In summary, ECG signal quality assessment is conducted by using unsupervised approach, supervised approach and their combination.…”

Section: Introductionmentioning

confidence: 99%

Improving Convolutional Neural Network with Recurrent Model and Generative Adversarial Networks for Imbalance Classification of Wearable ECG Signal Quality Assessment

Ma’sum

2022

J-Cosine

View full text Add to dashboard Cite

Signal quality assessment is important procedure to assess if a signal can be preprocessed into next step or not. In this study, we proposed an improving of convolutional neural network with recurrent model and generative adversarial network. The experiment shows that enhancing convolutional neural network by using recurrent model or data augmentation via GAN for wearable ECG signal quality assessment is feasible to try. Adding recurrent model to CNN increase its performance by approximately 3% margin in un-augmented dataset. Generating fake samples from “unacceptable” class can enhance CNN performance up to 2% margin. Combination of recurrent model and GAN still not yet matched found. However, It doesn’t mean that the combination is impossible to do.

show abstract