Pub Date : 2021-09-13DOI: 10.23919/cinc53138.2021.9662757
Long Chen, Zheheng Jiang, T. Almeida, F. Schlindwein, Jakevir S. Shoker, G. Ng, Huiyu Zhou, Xin Li
Automatic detection and classification of cardiac disorders play a critical role in the analysis of clinical electrocardiogram (ECG). Deep learning methods are effective for automated feature extraction and have shown promising results in ECG classification. In this work, we proposed a deep spatio-temporal ECG network (ST-ECGNet) to extract robust spatio-temporal features for detecting multiple cardiac disorders from the multi-lead ECG data. The proposed ST-ECGNet combines a Convolutional Neural Network (CNN) module for extracting local spatial features, an attention module for capturing global spatial features, and a Bi-directional Gated Recurrent Unit (Bi-GRU) module for extracting temporal features from ECG data. Specifically, the attention mechanism enables our deep learning architecture to focus on the most important and useful parts of the input to make more accurate predictions. In PhysioNet/Computing in Cardiology Challenge 2021, our entry was not officially ranked and scored on the test data of the Challenge, because our code was not successfully processed during the official phase and failed to run with errors.
心功能障碍的自动检测与分类在临床心电图分析中起着至关重要的作用。深度学习方法是一种有效的自动特征提取方法，在心电分类中显示出良好的效果。在这项工作中，我们提出了一个深度时空ECG网络(ST-ECGNet)来提取鲁棒的时空特征，用于从多导联ECG数据中检测多种心脏疾病。所提出的ST-ECGNet结合了卷积神经网络(CNN)模块用于提取局部空间特征，注意力模块用于捕获全局空间特征，双向门控循环单元(Bi-GRU)模块用于从心电数据中提取时间特征。具体来说，注意力机制使我们的深度学习架构能够专注于输入中最重要和最有用的部分，从而做出更准确的预测。在PhysioNet/Computing In Cardiology Challenge 2021中，我们的参赛作品没有在挑战赛的测试数据上得到正式的排名和评分，因为我们的代码在官方阶段没有被成功处理，并且错误地运行失败。
{"title":"Spatio-Temporal ECG Network for Detecting Cardiac Disorders from Multi-Lead ECGs","authors":"Long Chen, Zheheng Jiang, T. Almeida, F. Schlindwein, Jakevir S. Shoker, G. Ng, Huiyu Zhou, Xin Li","doi":"10.23919/cinc53138.2021.9662757","DOIUrl":"https://doi.org/10.23919/cinc53138.2021.9662757","url":null,"abstract":"Automatic detection and classification of cardiac disorders play a critical role in the analysis of clinical electrocardiogram (ECG). Deep learning methods are effective for automated feature extraction and have shown promising results in ECG classification. In this work, we proposed a deep spatio-temporal ECG network (ST-ECGNet) to extract robust spatio-temporal features for detecting multiple cardiac disorders from the multi-lead ECG data. The proposed ST-ECGNet combines a Convolutional Neural Network (CNN) module for extracting local spatial features, an attention module for capturing global spatial features, and a Bi-directional Gated Recurrent Unit (Bi-GRU) module for extracting temporal features from ECG data. Specifically, the attention mechanism enables our deep learning architecture to focus on the most important and useful parts of the input to make more accurate predictions. In PhysioNet/Computing in Cardiology Challenge 2021, our entry was not officially ranked and scored on the test data of the Challenge, because our code was not successfully processed during the official phase and failed to run with errors.","PeriodicalId":126746,"journal":{"name":"2021 Computing in Cardiology (CinC)","volume":"48 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-09-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"131329422","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2021-09-13DOI: 10.23919/cinc53138.2021.9662786
E. Sung, A. Prakosa, N. Trayanova
Repolarization heterogeneity contributes to ventricular tachycardia (VT) arrhythmogenesis but the impact of repolarization gradients on post-infarct VT dynamics is not well-characterized. The goal of our study is to assess the effects of repolarization gradients on post-infarct VT dynamics using patient-specific heart models. Baseline models were reconstructed along with the patient-specific scar and infarct border zone from imaging. Models with action potential duration (APD) gradients along apicobasal (AB) and transmural (TM) axes were also reconstructed. Rapid pacing was used to induce VTs. The resultant VT dynamics (inducibility, re-entry pathway, and the exit site) were assessed. Repolarization gradients did not impact VT inducibility but did alter both the reentry pathway and exit site location due to modulations in unidirectional conduction block. Both AB and TM APD gradients alone were also sufficient for inducing these changes in VT dynamics. Lastly, APD gradients revealed multiple distinct morphologies that used similar conducting channels in the patient-specific substrate. These results highlight how the interplay between repolarization gradients and the patient-specific substrate can have consequences on post-infarct VT dynamics.
{"title":"Repolarization Gradients Alter Post-infarct Ventricular Tachycardia Dynamics in Patient-Specific Computational Heart Models","authors":"E. Sung, A. Prakosa, N. Trayanova","doi":"10.23919/cinc53138.2021.9662786","DOIUrl":"https://doi.org/10.23919/cinc53138.2021.9662786","url":null,"abstract":"Repolarization heterogeneity contributes to ventricular tachycardia (VT) arrhythmogenesis but the impact of repolarization gradients on post-infarct VT dynamics is not well-characterized. The goal of our study is to assess the effects of repolarization gradients on post-infarct VT dynamics using patient-specific heart models. Baseline models were reconstructed along with the patient-specific scar and infarct border zone from imaging. Models with action potential duration (APD) gradients along apicobasal (AB) and transmural (TM) axes were also reconstructed. Rapid pacing was used to induce VTs. The resultant VT dynamics (inducibility, re-entry pathway, and the exit site) were assessed. Repolarization gradients did not impact VT inducibility but did alter both the reentry pathway and exit site location due to modulations in unidirectional conduction block. Both AB and TM APD gradients alone were also sufficient for inducing these changes in VT dynamics. Lastly, APD gradients revealed multiple distinct morphologies that used similar conducting channels in the patient-specific substrate. These results highlight how the interplay between repolarization gradients and the patient-specific substrate can have consequences on post-infarct VT dynamics.","PeriodicalId":126746,"journal":{"name":"2021 Computing in Cardiology (CinC)","volume":"306 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-09-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"123047309","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2021-09-13DOI: 10.23919/cinc53138.2021.9662770
J. Elliott, M. Belen, L. Mainardi, V. Corino, J. F. R. Matas
Improved understanding of the impact of variability on electrophysiological mechanisms is key to understanding the cause and development of cardiovascular disease. Recent studies suggest cellular variability could have an impact on electrophysiological behavior that homogeneous models are unable to capture. This study investigates the impact of cellular variability on conduction velocity and the depolarization and repolarization phases of the atria. Method: 10 Isolated tissue samples for each atrial region were calibrated for CV and later combined in a detailed anatomical atrial model. Variable models were compared with equivalent homogeneous models. Activation maps and APD maps were used for comparison. Results: In isolated tissue simulations, differences in tissue conductance (Gi) ranged between 5.5% reduction to 5.4% increase as a result of heterogeneity, despite differences in CV being <1%. Activation maps showed no significant differences between regionally homogeneous and heterogeneous atrial models. Repolarization across the atria differed significantly between regionally homogeneous and heterogeneous atrial models. Conclusion: Cellular variability has no significant impact on depolarization but significantly influences atrial repolarization. This could result in increased susceptibility to re-entries and atrial fibrillation.
{"title":"Impacts of Cellular Electrophysiological Variability on Conduction Velocity Within Isolated Tissue and Depolarization and Repolarization Across the Whole Atrial Model","authors":"J. Elliott, M. Belen, L. Mainardi, V. Corino, J. F. R. Matas","doi":"10.23919/cinc53138.2021.9662770","DOIUrl":"https://doi.org/10.23919/cinc53138.2021.9662770","url":null,"abstract":"Improved understanding of the impact of variability on electrophysiological mechanisms is key to understanding the cause and development of cardiovascular disease. Recent studies suggest cellular variability could have an impact on electrophysiological behavior that homogeneous models are unable to capture. This study investigates the impact of cellular variability on conduction velocity and the depolarization and repolarization phases of the atria. Method: 10 Isolated tissue samples for each atrial region were calibrated for CV and later combined in a detailed anatomical atrial model. Variable models were compared with equivalent homogeneous models. Activation maps and APD maps were used for comparison. Results: In isolated tissue simulations, differences in tissue conductance (Gi) ranged between 5.5% reduction to 5.4% increase as a result of heterogeneity, despite differences in CV being <1%. Activation maps showed no significant differences between regionally homogeneous and heterogeneous atrial models. Repolarization across the atria differed significantly between regionally homogeneous and heterogeneous atrial models. Conclusion: Cellular variability has no significant impact on depolarization but significantly influences atrial repolarization. This could result in increased susceptibility to re-entries and atrial fibrillation.","PeriodicalId":126746,"journal":{"name":"2021 Computing in Cardiology (CinC)","volume":"12 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-09-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"124457370","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
In the PhysioNet/Computing in Cardiology Challenge 2021, our team, DrCubic, develops a novel approach to classify cardiac abnormalities using reduced-lead ECG recordings. In our approach, we incorporate peak detection as a self-supervised auxiliary task. We build the model based on SE-ResNet, and integrate models of different input lengths and sampling rates. Inspired by last year's challenge results, we investigate various settings and techniques, and select the best ones, considering the intra-source performance and inter-source generalization simultaneously. Our classifiers receive scores of 0.49, 0.50, 0.50, 0.51, and 0.48 (ranked 9th, 8th, 7th, 5th, and 9th out of 39 scored teams) for the 12 -lead, 6-lead, 4-lead, 3-lead, and 2 -lead versions of the hidden test sets with the Challenge evaluation metric.
在PhysioNet/Computing In Cardiology Challenge 2021中，我们的团队DrCubic开发了一种使用降导联心电图记录对心脏异常进行分类的新方法。在我们的方法中，我们将峰值检测作为自监督辅助任务。我们基于SE-ResNet建立了模型，并整合了不同输入长度和采样率的模型。受去年挑战赛结果的启发，我们研究了各种设置和技术，并选择了最佳设置和技术，同时考虑了源内性能和源间泛化。我们的分类器收到的分数分别为0.49,0.50,0.50,0.51和0.48(在39个得分的团队中排名第9，第8，第7，第5和第9)，用于12领先，6领先，4领先，3领先和2领先版本的隐藏测试集与挑战评估指标。
{"title":"Towards Generalization of Cardiac Abnormality Classification Using ECG Signal","authors":"Xiaoyu Li, Chen Li, Xian Xu, Yuhua Wei, Jishang Wei, Yuyao Sun, B. Qian, Xiao Xu","doi":"10.23919/cinc53138.2021.9662822","DOIUrl":"https://doi.org/10.23919/cinc53138.2021.9662822","url":null,"abstract":"In the PhysioNet/Computing in Cardiology Challenge 2021, our team, DrCubic, develops a novel approach to classify cardiac abnormalities using reduced-lead ECG recordings. In our approach, we incorporate peak detection as a self-supervised auxiliary task. We build the model based on SE-ResNet, and integrate models of different input lengths and sampling rates. Inspired by last year's challenge results, we investigate various settings and techniques, and select the best ones, considering the intra-source performance and inter-source generalization simultaneously. Our classifiers receive scores of 0.49, 0.50, 0.50, 0.51, and 0.48 (ranked 9th, 8th, 7th, 5th, and 9th out of 39 scored teams) for the 12 -lead, 6-lead, 4-lead, 3-lead, and 2 -lead versions of the hidden test sets with the Challenge evaluation metric.","PeriodicalId":126746,"journal":{"name":"2021 Computing in Cardiology (CinC)","volume":"22 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-09-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"128992671","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2021-09-13DOI: 10.23919/cinc53138.2021.9662651
Narimane Gassa, Benjamin Sacristan, N. Zemzemi, M. Laborde, Juan Garrido Oliver, Clara Matencio Perabla, G. Jiménez-Pérez, O. Camara, S. Ploux, M. Strik, P. Bordachar, R. Dubois
The objective of this work was to benchmark different deep learning architectures for noise detection against cardiac arrhythmia episodes recorded by pacemakers and implantable cardioverter-defibrillators (PM/ICDs) and transmitted for remote monitoring. Up to now, most signal processing from ICD data has been based on classical hand-crafted algorithms, not AI or DL-based ones. The database consist of PM/ICD data from 805 patients representing a total of 10471 recordings from three different channels: the right ventricular (RV), the right atria (RA), and the shock channel. Four deep learning approaches were trained and optimized to classify PM/ICDs' records as actual ventricular signal vs noise episodes. We evaluated the performance of the different models using the F2 score. Results show that the use of 2D representations of 1D signals led to better performances than the direct use of 1D signals, suggesting that the detection of noise takes advantage of a spectral decomposition of the signal, which remains to be confirmed in other contexts. This study proposes deep learning approaches for the analysis of remote monitoring recordings from PM/ICDs. The detection of noise allows efficient management of this large daily flow of data.
{"title":"Benchmark of deep learning algorithms for the automatic screening in electrocardiograms transmitted by implantable cardiac devices","authors":"Narimane Gassa, Benjamin Sacristan, N. Zemzemi, M. Laborde, Juan Garrido Oliver, Clara Matencio Perabla, G. Jiménez-Pérez, O. Camara, S. Ploux, M. Strik, P. Bordachar, R. Dubois","doi":"10.23919/cinc53138.2021.9662651","DOIUrl":"https://doi.org/10.23919/cinc53138.2021.9662651","url":null,"abstract":"The objective of this work was to benchmark different deep learning architectures for noise detection against cardiac arrhythmia episodes recorded by pacemakers and implantable cardioverter-defibrillators (PM/ICDs) and transmitted for remote monitoring. Up to now, most signal processing from ICD data has been based on classical hand-crafted algorithms, not AI or DL-based ones. The database consist of PM/ICD data from 805 patients representing a total of 10471 recordings from three different channels: the right ventricular (RV), the right atria (RA), and the shock channel. Four deep learning approaches were trained and optimized to classify PM/ICDs' records as actual ventricular signal vs noise episodes. We evaluated the performance of the different models using the F2 score. Results show that the use of 2D representations of 1D signals led to better performances than the direct use of 1D signals, suggesting that the detection of noise takes advantage of a spectral decomposition of the signal, which remains to be confirmed in other contexts. This study proposes deep learning approaches for the analysis of remote monitoring recordings from PM/ICDs. The detection of noise allows efficient management of this large daily flow of data.","PeriodicalId":126746,"journal":{"name":"2021 Computing in Cardiology (CinC)","volume":"12 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-09-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"129154275","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2021-09-13DOI: 10.23919/cinc53138.2021.9662756
M. Hansen, B. Grønfeldt, Tue Rømer, Mathilde Fogelstrøm, Kasper Sørensen, S. Schmidt, J. Helge
Introduction: Assessment of maximal oxygen consumption (VO2max) is an important clinical tool when examining both healthy and unhealthy populations, as a low VO2max is associated with cardiovascular disease and all-cause mortality. Aim: This study investigated the accuracy of a non-exercise test for assessment of VO2max using seismocardiography (SCG). Methods: 97 participants (20–45 years, 50 males) underwent a nonexercise test using SCG at rest in the supine position (SCG VO2max) and a graded exercise test to voluntary exhaustion on a cycle ergometer with indirect calorimetry (IC VO2max). An interim analysis was applied after 50 participants had completed testing (SCG VO2max 1.0) allowing for the algorithm to be modified (SCG VO2max 2.1). Results: SCG VO2max 2.1 ($n=47$, test set) estimation was $3.5 pm 1.8 mlcdot min^{-1}cdot kg^{-1} (p < 0.001)$ lower compared to IC VO2max, with a Pearson correlation of $r=0.65 (p < 0.0001)$ and a standard error of estimate of 7.1 ml·min−1 ·kg−1. The coefficient of variation between tests was $8 pm 1%$. Conclusion: The accuracy of VO2max assessment using SCG requires further optimization prior to clinical application, as SCG VO2max was systematically lower than IC VO2max, and only a moderate correlation together with considerable variation were observed between tests.
{"title":"Determination of Maximal Oxygen Uptake Using Seismocardiography at Rest","authors":"M. Hansen, B. Grønfeldt, Tue Rømer, Mathilde Fogelstrøm, Kasper Sørensen, S. Schmidt, J. Helge","doi":"10.23919/cinc53138.2021.9662756","DOIUrl":"https://doi.org/10.23919/cinc53138.2021.9662756","url":null,"abstract":"Introduction: Assessment of maximal oxygen consumption (VO<inf>2</inf>max) is an important clinical tool when examining both healthy and unhealthy populations, as a low VO<inf>2</inf>max is associated with cardiovascular disease and all-cause mortality. Aim: This study investigated the accuracy of a non-exercise test for assessment of VO<inf>2</inf>max using seismocardiography (SCG). Methods: 97 participants (20–45 years, 50 males) underwent a nonexercise test using SCG at rest in the supine position (SCG VO<inf>2</inf>max) and a graded exercise test to voluntary exhaustion on a cycle ergometer with indirect calorimetry (IC VO<inf>2</inf>max). An interim analysis was applied after 50 participants had completed testing (SCG VO<inf>2</inf>max 1.0) allowing for the algorithm to be modified (SCG VO<inf>2</inf>max 2.1). Results: SCG VO<inf>2</inf>max 2.1 (<tex>$n=47$</tex>, test set) estimation was <tex>$3.5 pm 1.8 mlcdot min^{-1}cdot kg^{-1} (p < 0.001)$</tex> lower compared to IC VO<inf>2</inf>max, with a Pearson correlation of <tex>$r=0.65 (p < 0.0001)$</tex> and a standard error of estimate of 7.1 ml·min<sup>−1</sup> ·kg<sup>−1</sup>. The coefficient of variation between tests was <tex>$8 pm 1%$</tex>. Conclusion: The accuracy of VO<inf>2</inf>max assessment using SCG requires further optimization prior to clinical application, as SCG VO<inf>2</inf>max was systematically lower than IC VO<inf>2</inf>max, and only a moderate correlation together with considerable variation were observed between tests.","PeriodicalId":126746,"journal":{"name":"2021 Computing in Cardiology (CinC)","volume":"48 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-09-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"128834014","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2021-09-13DOI: 10.23919/cinc53138.2021.9662932
Martin Schmidt, Filip Karisik, S. Zaunseder, A. Linke, H. Malberg, M. Baumert
To investigate the predictive value of Ventricular repo-larization variability (VRV) in patients with nonischemic dilated cardiomyopathy, we analyzed the Defibrillator in Non-Ischemic Cardiomyopathy Treatment Evaluation trial (DEFINITE). The Telemetric and Holter ECG Warehouse (THEW) data set E-HOL-03-0401-017 comprises 393 recordings from 236 patients. All patients had a left ventricular ejection fraction $< 36$ % and were randomized to receiving standard medical therapy with or without an ICD. 24h-Holter 3-lead (Frank lead system) ECGs were performed at enrollment and after up to 5 years' follow-up. The all-cause mortality during the follow-up period was 4.8 %. We analyzed three-dimensional variability of the T-loop and QT interval variability on a single lead basis by employing three-dimensional signals adaptation and two-dimensional signal warping, respectively, to quantify VRV. To assess the predictive value of VRV parameters, Kaplan-Meier survival curves of baseline Holter ECGs were calculated. Our results showed significant association to survival ( $P < 0.01$ by the log-rank test) for T wave amplitude corrected QT interval variability index (cQTVi) on single lead basis. Low cQTVi group showed no mortality for the entire observation period. We found no associations between cQTVi groups and patient-specific parameters.
{"title":"Evaluation of Ventricular Repolarization Variability in Patients With Nonischemic Dilated Cardiomyopathy From Vectorcardiography","authors":"Martin Schmidt, Filip Karisik, S. Zaunseder, A. Linke, H. Malberg, M. Baumert","doi":"10.23919/cinc53138.2021.9662932","DOIUrl":"https://doi.org/10.23919/cinc53138.2021.9662932","url":null,"abstract":"To investigate the predictive value of Ventricular repo-larization variability (VRV) in patients with nonischemic dilated cardiomyopathy, we analyzed the Defibrillator in Non-Ischemic Cardiomyopathy Treatment Evaluation trial (DEFINITE). The Telemetric and Holter ECG Warehouse (THEW) data set E-HOL-03-0401-017 comprises 393 recordings from 236 patients. All patients had a left ventricular ejection fraction $< 36$ % and were randomized to receiving standard medical therapy with or without an ICD. 24h-Holter 3-lead (Frank lead system) ECGs were performed at enrollment and after up to 5 years' follow-up. The all-cause mortality during the follow-up period was 4.8 %. We analyzed three-dimensional variability of the T-loop and QT interval variability on a single lead basis by employing three-dimensional signals adaptation and two-dimensional signal warping, respectively, to quantify VRV. To assess the predictive value of VRV parameters, Kaplan-Meier survival curves of baseline Holter ECGs were calculated. Our results showed significant association to survival ( $P < 0.01$ by the log-rank test) for T wave amplitude corrected QT interval variability index (cQTVi) on single lead basis. Low cQTVi group showed no mortality for the entire observation period. We found no associations between cQTVi groups and patient-specific parameters.","PeriodicalId":126746,"journal":{"name":"2021 Computing in Cardiology (CinC)","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-09-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"128581088","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2021-09-13DOI: 10.23919/cinc53138.2021.9662691
John Wang
Real-time ST-segment monitoring for ischemia detection was introduced for clinical use in the 80s. To overcome the earlier systems' limitation on the number of leads monitored, systems that support continuous 12-lead ECG acquisition were developed. Derived 12-lead ECGs from 5-wire and 6-wire lead sets were also developed when direct 12 -lead acquisition was not practical. Several innovative graphical solutions were developed to manage the large amount of date from continuous 12-lead ST monitoring, including ST Map for better visual tracking of ST measurements, STEMI Map for more accurate tracking of STEMI criteria, and ST Topology for more efficient ST trending review. To further improve the accuracy of acute ischemia/infraction detection, two advanced 12-lead based lead derivation methods are being developed. The vessel-specific leads (VSLs) method measures ST elevation from three optimal leads, calculated from the 12-lead ECG, for detecting ST-segment deviation during coronary occlusion. The computed electrocardiographic imaging (CEI) method presents a bulls-eye polar plot of the heart surface potentials based on inverse calculation from the body-surface potential mapping derived from the 12-lead ECG. Early results show that these methods could be a useful clinical decision support tool for improving the accuracy of ECG-based triage of chest-pain patients.
{"title":"Advances in ECG-Based Cardiac Ischemia Monitoring - A Review","authors":"John Wang","doi":"10.23919/cinc53138.2021.9662691","DOIUrl":"https://doi.org/10.23919/cinc53138.2021.9662691","url":null,"abstract":"Real-time ST-segment monitoring for ischemia detection was introduced for clinical use in the 80s. To overcome the earlier systems' limitation on the number of leads monitored, systems that support continuous 12-lead ECG acquisition were developed. Derived 12-lead ECGs from 5-wire and 6-wire lead sets were also developed when direct 12 -lead acquisition was not practical. Several innovative graphical solutions were developed to manage the large amount of date from continuous 12-lead ST monitoring, including ST Map for better visual tracking of ST measurements, STEMI Map for more accurate tracking of STEMI criteria, and ST Topology for more efficient ST trending review. To further improve the accuracy of acute ischemia/infraction detection, two advanced 12-lead based lead derivation methods are being developed. The vessel-specific leads (VSLs) method measures ST elevation from three optimal leads, calculated from the 12-lead ECG, for detecting ST-segment deviation during coronary occlusion. The computed electrocardiographic imaging (CEI) method presents a bulls-eye polar plot of the heart surface potentials based on inverse calculation from the body-surface potential mapping derived from the 12-lead ECG. Early results show that these methods could be a useful clinical decision support tool for improving the accuracy of ECG-based triage of chest-pain patients.","PeriodicalId":126746,"journal":{"name":"2021 Computing in Cardiology (CinC)","volume":"42 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-09-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"116410044","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2021-09-13DOI: 10.23919/cinc53138.2021.9662929
S. Krivenko, A. Pulavskyi, L. Kryvenko, O. Krylova, Sergey A. Krivenko
We have developed the XGBoost model to identify 27 heart pathologies within the challenge Will Two Do? Varying Dimensions in Electrocardiography: The PhysioNet/ Computing in Cardiology Challenge 2021. The technical part included several stages. At the first stage, the ECG was cut off to 10 seconds. At the second stage, resampling to frequencies 125 and 500 Hz was carried out and filtering in the 0.5-45 Hz bands. At the third stage, the features of HRV and symbolic dynamics were extracted from the signal with a sampling rate of 125 Hz. The melspectrograms were calculated based on a signal with a sampling frequency of 500 Hz. Then the features calculated for each lead were concatenated to obtain the final vector of features. We were faced with the task of constructing 27 independent binary classifiers, each of which defines a certain pathology. The fourth important step was to build balanced datasets for the algorithm. For the robustness of the models, the control groups for each contained almost all pathologies presented in the databases, except target disease. Our team Sunset scored 0.22, 0.21, 0.22, 0.21, 0.20 for the 12-lead, 6-lead, 4-lead, 3-lead, and 2-lead models, respectively, ranking 32 out of 39 teams for the first four lead combinations and 31 out of 39 teams for the last.
我们开发了XGBoost模型，以识别挑战Will Two Do?中的27种心脏疾病。心电图的不同维度:2021年心脏病学挑战中的物理网络/计算。技术部分包括几个阶段。在第一阶段，心电图被切断到10秒。在第二阶段，对频率125和500 Hz进行重采样，并在0.5-45 Hz频段进行滤波。第三阶段，以125 Hz的采样率提取信号的HRV特征和符号动力学特征。以采样频率为500hz的信号为基础，计算melogram。然后将每条引线计算的特征进行连接，得到最终的特征向量。我们面临的任务是构建27个独立的二元分类器，每个分类器定义一个特定的病理。第四步是为算法建立平衡的数据集。为了模型的稳健性，除了目标疾病外，每个对照组几乎包含数据库中提供的所有病理。我们队在12领先、6领先、4领先、3领先、2领先模式中得分分别为0.22、0.21、0.22、0.21、0.20，在前4种领先组合的39支队伍中排名第32位，在后39支队伍中排名第31位。
{"title":"Using Mel-Frequency Cepstrum and Amplitude-Time Heart Variability as XGBoost Handcrafted Features for Heart Disease Detection","authors":"S. Krivenko, A. Pulavskyi, L. Kryvenko, O. Krylova, Sergey A. Krivenko","doi":"10.23919/cinc53138.2021.9662929","DOIUrl":"https://doi.org/10.23919/cinc53138.2021.9662929","url":null,"abstract":"We have developed the XGBoost model to identify 27 heart pathologies within the challenge Will Two Do? Varying Dimensions in Electrocardiography: The PhysioNet/ Computing in Cardiology Challenge 2021. The technical part included several stages. At the first stage, the ECG was cut off to 10 seconds. At the second stage, resampling to frequencies 125 and 500 Hz was carried out and filtering in the 0.5-45 Hz bands. At the third stage, the features of HRV and symbolic dynamics were extracted from the signal with a sampling rate of 125 Hz. The melspectrograms were calculated based on a signal with a sampling frequency of 500 Hz. Then the features calculated for each lead were concatenated to obtain the final vector of features. We were faced with the task of constructing 27 independent binary classifiers, each of which defines a certain pathology. The fourth important step was to build balanced datasets for the algorithm. For the robustness of the models, the control groups for each contained almost all pathologies presented in the databases, except target disease. Our team Sunset scored 0.22, 0.21, 0.22, 0.21, 0.20 for the 12-lead, 6-lead, 4-lead, 3-lead, and 2-lead models, respectively, ranking 32 out of 39 teams for the first four lead combinations and 31 out of 39 teams for the last.","PeriodicalId":126746,"journal":{"name":"2021 Computing in Cardiology (CinC)","volume":"26 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-09-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"116756399","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2021-09-13DOI: 10.23919/cinc53138.2021.9662912
K. Čurila, P. Jurák, P. Leinveber, R. Smíšek, P. Stros, F. Plesinger, I. Viscor, V. Vondra, J. Mizner, O. Sussenbek, L. Znojilova, J. Karch, M. Susankova, J. Halámek, F. Prinzen
Background: Permanent cardiac pacing can cause heart failure, with the ventricular dyssynchrony being identified as the main cause for its development. Method: His bundle pacing (HBp), left bundle branch pacing (LBBp), and left ventricular myocardial septal pacing (LVSP) were introduced recently. Their impact on ventricular dyssynchrony was not known. We used ultra-high-frequency ECG (UHF-ECG) to compare ventricular depolarization in these pacing techniques. Results: We showed the nonselective HB pacing produces the same pattern of UHF-ECG ventricular depolarization as selective HB pacing. Next, we showed the nonselective His bundle pacing in the area below the tricuspid valve has the best interventricular synchrony from all other RV pacing locations with myocardial capture. We also compared UHF-ECG-derived parameters of ventricular depolarization during HBp, LBBp, and LVSP and we showed that both pacing types from the left septal area are less physiological than nsHBp. Conclusion: UHF-ECG is an effective tool that can be used in clinical practice to assess the electrical dyssynchrony caused by cardiac pacing. Furthermore, its real-time implementation allows recognizing between physiological vs. non-physiological pacing during an implant procedure.
{"title":"Physiological versus non-physiological cardiac pacing as assessed by Ultra-high-frequency electrocardiography","authors":"K. Čurila, P. Jurák, P. Leinveber, R. Smíšek, P. Stros, F. Plesinger, I. Viscor, V. Vondra, J. Mizner, O. Sussenbek, L. Znojilova, J. Karch, M. Susankova, J. Halámek, F. Prinzen","doi":"10.23919/cinc53138.2021.9662912","DOIUrl":"https://doi.org/10.23919/cinc53138.2021.9662912","url":null,"abstract":"Background: Permanent cardiac pacing can cause heart failure, with the ventricular dyssynchrony being identified as the main cause for its development. Method: His bundle pacing (HBp), left bundle branch pacing (LBBp), and left ventricular myocardial septal pacing (LVSP) were introduced recently. Their impact on ventricular dyssynchrony was not known. We used ultra-high-frequency ECG (UHF-ECG) to compare ventricular depolarization in these pacing techniques. Results: We showed the nonselective HB pacing produces the same pattern of UHF-ECG ventricular depolarization as selective HB pacing. Next, we showed the nonselective His bundle pacing in the area below the tricuspid valve has the best interventricular synchrony from all other RV pacing locations with myocardial capture. We also compared UHF-ECG-derived parameters of ventricular depolarization during HBp, LBBp, and LVSP and we showed that both pacing types from the left septal area are less physiological than nsHBp. Conclusion: UHF-ECG is an effective tool that can be used in clinical practice to assess the electrical dyssynchrony caused by cardiac pacing. Furthermore, its real-time implementation allows recognizing between physiological vs. non-physiological pacing during an implant procedure.","PeriodicalId":126746,"journal":{"name":"2021 Computing in Cardiology (CinC)","volume":"17 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-09-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"116813580","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：