Pub Date : 2019-09-01DOI: 10.23919/CinC49843.2019.9005810
M. Alfaras, Rui Varandas, H. Gamboa
Sepsis is a life threatening condition that can be treated if detected early. This paper presents a study of the application of a Ring Topology Echo State Network (ESN) algorithm to a sepsis prediction task based on ICU records. The implemented algorithm is compared with commonly used classifiers and a combination of both approaches. Finally, we address how different causal strategies on filling missing record values affected the final classification performances. Having a dataset with a limited number of time entries per patient, the utility score U = 0.188 obtained (team 51: PLUX) suggests that further research is needed in order for the ESN to capture the temporal dynamics of the problem at hand.
{"title":"Ring-Topology Echo State Networks for ICU Sepsis Classification","authors":"M. Alfaras, Rui Varandas, H. Gamboa","doi":"10.23919/CinC49843.2019.9005810","DOIUrl":"https://doi.org/10.23919/CinC49843.2019.9005810","url":null,"abstract":"Sepsis is a life threatening condition that can be treated if detected early. This paper presents a study of the application of a Ring Topology Echo State Network (ESN) algorithm to a sepsis prediction task based on ICU records. The implemented algorithm is compared with commonly used classifiers and a combination of both approaches. Finally, we address how different causal strategies on filling missing record values affected the final classification performances. Having a dataset with a limited number of time entries per patient, the utility score U = 0.188 obtained (team 51: PLUX) suggests that further research is needed in order for the ESN to capture the temporal dynamics of the problem at hand.","PeriodicalId":6697,"journal":{"name":"2019 Computing in Cardiology (CinC)","volume":"68 1","pages":"Page 1-Page 4"},"PeriodicalIF":0.0,"publicationDate":"2019-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"85397757","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 : 2019-09-01DOI: 10.23919/CinC49843.2019.9005678
G. Santarelli, Roberta Ciccotelli, G. Molon, F. Zanon, A. Corzani, A. Rossillo, M. Biffi, G. Zanotto, L. Lanzoni, S. Severi, C. Tomasi, C. Corsi
Cardiac resynchronization therapy (CRT) is an effective treatment for chronic symptomatic systolic heart failure with cardiac dyssynchrony, but about 1/3 of patients do not respond favorably to the therapy. We hypothesized that acute modifications of the coronary sinus (CS) pacing cathode movements induced by biventricular pacing may be related to resynchronization process and consequently may carry predictive power on CRT response. A method for the 3D reconstruction of CS lead’s pacing cathode trajectory (3DTJ) throughout a cardiac cycle showed that trajectory’s geometry suddenly changed in responders (R) upon starting of biventricular pacing, becoming less eccentric and more multi-directional. Our multicenter observational study aimed at evaluating the clinical value of 3DTJ. Out of 119 patients enrolled, 50 have ended follow-up and have been analyzed. Concordance between 3DTJ metrics and response was 82% overall (41/50), 91% in R (31/34), 62% in NR (10/16). The proposed 3DTJ metric showed high sensitivity (91%) with specificity=62%; PPV=84%, NPV=77%. From our data, 3DTJ seems a promising tool to acutely predict CS pacing site-specific response to CRT. Its investigational use as an intra-operatory, real-time guidance for selecting LV pacing sites may open new opportunities for CRT patients’ selection and therapy delivery.
{"title":"Evaluation of Short-Term Pacing Effect to Predict Long-Term Response to Cardiac Resynchronization Therapy: the TRAJECTORIES Study","authors":"G. Santarelli, Roberta Ciccotelli, G. Molon, F. Zanon, A. Corzani, A. Rossillo, M. Biffi, G. Zanotto, L. Lanzoni, S. Severi, C. Tomasi, C. Corsi","doi":"10.23919/CinC49843.2019.9005678","DOIUrl":"https://doi.org/10.23919/CinC49843.2019.9005678","url":null,"abstract":"Cardiac resynchronization therapy (CRT) is an effective treatment for chronic symptomatic systolic heart failure with cardiac dyssynchrony, but about 1/3 of patients do not respond favorably to the therapy. We hypothesized that acute modifications of the coronary sinus (CS) pacing cathode movements induced by biventricular pacing may be related to resynchronization process and consequently may carry predictive power on CRT response. A method for the 3D reconstruction of CS lead’s pacing cathode trajectory (3DTJ) throughout a cardiac cycle showed that trajectory’s geometry suddenly changed in responders (R) upon starting of biventricular pacing, becoming less eccentric and more multi-directional. Our multicenter observational study aimed at evaluating the clinical value of 3DTJ. Out of 119 patients enrolled, 50 have ended follow-up and have been analyzed. Concordance between 3DTJ metrics and response was 82% overall (41/50), 91% in R (31/34), 62% in NR (10/16). The proposed 3DTJ metric showed high sensitivity (91%) with specificity=62%; PPV=84%, NPV=77%. From our data, 3DTJ seems a promising tool to acutely predict CS pacing site-specific response to CRT. Its investigational use as an intra-operatory, real-time guidance for selecting LV pacing sites may open new opportunities for CRT patients’ selection and therapy delivery.","PeriodicalId":6697,"journal":{"name":"2019 Computing in Cardiology (CinC)","volume":"451 1","pages":"Page 1-Page 4"},"PeriodicalIF":0.0,"publicationDate":"2019-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"79702803","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 : 2019-09-01DOI: 10.23919/CinC49843.2019.9005768
P. Macfarlane, S. Latif, B. Devine
There has been interest relating to automated analysis of a lead I ECG to detect cardiac arrhythmias. Little interest has been shown in the accuracy of using lead I as opposed to 6 limb leads or the full 12 lead ECG. The aim of this small study was to assess the efficacy of using only lead I but also to look at the effect of analysing a single 30s recording as a continuous recording versus five 10s overlapping recordings constituting a 30s record.One hundred 10s digital 12 lead ECGs with atrial fibrillation (AF) were used. Chest leads were removed and the 6 limb leads then used for analysis of rhythm. Similarly, lead I alone was used. Separately 100 single lead I ECGs classified as AF in the PhysioNet 2017 database were analysed, both as single 30s recordings and as five 10s ECGs commencing at 0, 5, 10, 15 and 20s from the start of the recording. An algorithm made the diagnosis from 5 reports. All analyses were made with the Glasgow Program. For the 10s 12 lead ECGs, 96% were reported as AF using 6 limb leads and 93% using lead I. For the 30s recordings, 92% were reported as AF using a single 30s analysis and 91% as AF using the five ECGs.In conclusion, one lead and 6 leads are not as sensitive as 12 leads in detecting AF, while five 10s reports combined are no more sensitive than a single 30s report though more specific.
{"title":"Factors Influencing Automated Limited Lead Detection of Atrial Fibrillation","authors":"P. Macfarlane, S. Latif, B. Devine","doi":"10.23919/CinC49843.2019.9005768","DOIUrl":"https://doi.org/10.23919/CinC49843.2019.9005768","url":null,"abstract":"There has been interest relating to automated analysis of a lead I ECG to detect cardiac arrhythmias. Little interest has been shown in the accuracy of using lead I as opposed to 6 limb leads or the full 12 lead ECG. The aim of this small study was to assess the efficacy of using only lead I but also to look at the effect of analysing a single 30s recording as a continuous recording versus five 10s overlapping recordings constituting a 30s record.One hundred 10s digital 12 lead ECGs with atrial fibrillation (AF) were used. Chest leads were removed and the 6 limb leads then used for analysis of rhythm. Similarly, lead I alone was used. Separately 100 single lead I ECGs classified as AF in the PhysioNet 2017 database were analysed, both as single 30s recordings and as five 10s ECGs commencing at 0, 5, 10, 15 and 20s from the start of the recording. An algorithm made the diagnosis from 5 reports. All analyses were made with the Glasgow Program. For the 10s 12 lead ECGs, 96% were reported as AF using 6 limb leads and 93% using lead I. For the 30s recordings, 92% were reported as AF using a single 30s analysis and 91% as AF using the five ECGs.In conclusion, one lead and 6 leads are not as sensitive as 12 leads in detecting AF, while five 10s reports combined are no more sensitive than a single 30s report though more specific.","PeriodicalId":6697,"journal":{"name":"2019 Computing in Cardiology (CinC)","volume":"38 1","pages":"Page 1-Page 4"},"PeriodicalIF":0.0,"publicationDate":"2019-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"83862380","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 : 2019-09-01DOI: 10.23919/CinC49843.2019.9005552
Yu-He Zhang, S. Babaeizadeh
Without reference annotation, statistical metrics such as sensitivity and positive predictive value (PPV) cannot be calculated. Annotating a large ECG database may not be feasible, hence, the interest in developing an evaluation tool that does not require reference annotation. We developed a tool for evaluating key performance attributes (KPA) including arrhythmia detection, heart rate, ST value, and noise tolerance. The tool has three layers of KPA graphics. The top layer includes interactive distribution graphs of the KPA values for aggregated results for the entire database. From this top layer the user can select an individual record to launch interactive trending graphs that display the KPA values, or their discrepancies, for a time span on that particular record. From this second layer the user can identify any KPA value of interest (e.g., a specific arrhythmia label) to view the underlying ECG waveform. Navigating through these three layers, the user is able to quickly confirm the validity of KPA reported by the algorithm. We modified the noise tolerance of an exercise ECG arrhythmia algorithm. Then used this tool to visually verify the resulting improvement on the Telemetric and Holter ECG Warehouse (THEW) stress database E-OTH-12-0927-015. We confirmed the visual verification of improvement by manually annotating a small subset of records in this database.
{"title":"A Graphical Evaluation Tool to Utilize ECG Data Without Reference Annotation","authors":"Yu-He Zhang, S. Babaeizadeh","doi":"10.23919/CinC49843.2019.9005552","DOIUrl":"https://doi.org/10.23919/CinC49843.2019.9005552","url":null,"abstract":"Without reference annotation, statistical metrics such as sensitivity and positive predictive value (PPV) cannot be calculated. Annotating a large ECG database may not be feasible, hence, the interest in developing an evaluation tool that does not require reference annotation. We developed a tool for evaluating key performance attributes (KPA) including arrhythmia detection, heart rate, ST value, and noise tolerance. The tool has three layers of KPA graphics. The top layer includes interactive distribution graphs of the KPA values for aggregated results for the entire database. From this top layer the user can select an individual record to launch interactive trending graphs that display the KPA values, or their discrepancies, for a time span on that particular record. From this second layer the user can identify any KPA value of interest (e.g., a specific arrhythmia label) to view the underlying ECG waveform. Navigating through these three layers, the user is able to quickly confirm the validity of KPA reported by the algorithm. We modified the noise tolerance of an exercise ECG arrhythmia algorithm. Then used this tool to visually verify the resulting improvement on the Telemetric and Holter ECG Warehouse (THEW) stress database E-OTH-12-0927-015. We confirmed the visual verification of improvement by manually annotating a small subset of records in this database.","PeriodicalId":6697,"journal":{"name":"2019 Computing in Cardiology (CinC)","volume":"27 1","pages":"1-4"},"PeriodicalIF":0.0,"publicationDate":"2019-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"81008391","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 : 2019-09-01DOI: 10.23919/CinC49843.2019.9005565
Luan Tran, M. Nguyen, C. Shahabi
As part of the PhysioNet/Computing in Cardiology Challenge 2019, we propose a neural network called AEC-Net to early detect sepsis based on physiological data. AEC-Net consists of two main components: 1) an Auto Encoder for dimension reduction and feature extraction, and 2) a Fully Connected Neural Network (FCNN) taking the extracted features by the Auto Encoder as the input and generating prediction of sepsis as output. The losses of both the Auto Encoder and FCNN are minimized concurrently. This concurrent optimization helps AEC-Net to have a better generalization and the extracted features by Auto Encoder to be more relevant to the classification problem. Finally, we propose an ensemble method of AEC-Net, Random Forest and Gradient Boosting Decision Trees to achieve a better prediction.We train our proposed models using data from 40336 patients with 40 physiological features ranging from 8 to 336 hours. Our team Infolab USC evaluated Ensemble with the hidden full test set of the Physionet Challenge 2019, and achieved a Utility score of 0.284 and 24th place in the challenge.
{"title":"Representation Learning for Early Sepsis Prediction","authors":"Luan Tran, M. Nguyen, C. Shahabi","doi":"10.23919/CinC49843.2019.9005565","DOIUrl":"https://doi.org/10.23919/CinC49843.2019.9005565","url":null,"abstract":"As part of the PhysioNet/Computing in Cardiology Challenge 2019, we propose a neural network called AEC-Net to early detect sepsis based on physiological data. AEC-Net consists of two main components: 1) an Auto Encoder for dimension reduction and feature extraction, and 2) a Fully Connected Neural Network (FCNN) taking the extracted features by the Auto Encoder as the input and generating prediction of sepsis as output. The losses of both the Auto Encoder and FCNN are minimized concurrently. This concurrent optimization helps AEC-Net to have a better generalization and the extracted features by Auto Encoder to be more relevant to the classification problem. Finally, we propose an ensemble method of AEC-Net, Random Forest and Gradient Boosting Decision Trees to achieve a better prediction.We train our proposed models using data from 40336 patients with 40 physiological features ranging from 8 to 336 hours. Our team Infolab USC evaluated Ensemble with the hidden full test set of the Physionet Challenge 2019, and achieved a Utility score of 0.284 and 24th place in the challenge.","PeriodicalId":6697,"journal":{"name":"2019 Computing in Cardiology (CinC)","volume":"75 1","pages":"1-4"},"PeriodicalIF":0.0,"publicationDate":"2019-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"89530540","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 : 2019-09-01DOI: 10.23919/CinC49843.2019.9005605
Frank van Rosmalen, L. Pison, T. Delhaas, H. Crijns, S. Zeemering, U. Schotten
Background: Pulmonary vein isolation (PVI) as treatment for atrial fibrillation (AF) is not effective in up to 60% of patients with persistent AF; AF drivers outside of the pulmonary veins can contribute to AF recurrences after PVI. In this study we explored the potential use of local conduction velocity (CV) during pacing as a marker of left atrial (LA) substrate complexity.Methods: LA activation times were recorded for 7 AF patients during coronary sinus (CS) pacing before PVI using a Pentaray catheter. Activation times were relative to the CS pacing spike. LA activation locations were triangularized to calculate CV: the local direction and speed of the activation wave front. CV was quantified by the total CV distribution.Results: A mean of 1622 CVs were calculated per patient. Distribution of CVs showed a similar morphology, with median CVs in the range [0.26, 0.36] and interquartile ranges in the range [0.29, 0.39].Conclusion: This study shows that although it is feasible to calculate CVs based on sequential CARTO mapping of the LA during CS pacing, the resulting distribution of CVs using this procedure is not necessarily able to identify substrate complexity because of the large similarity between distributions and the relatively small differences in medians.
{"title":"Local Atrial Conduction Velocity During Pacing as Indication of Atrial Fibrillation Substrate Complexity","authors":"Frank van Rosmalen, L. Pison, T. Delhaas, H. Crijns, S. Zeemering, U. Schotten","doi":"10.23919/CinC49843.2019.9005605","DOIUrl":"https://doi.org/10.23919/CinC49843.2019.9005605","url":null,"abstract":"Background: Pulmonary vein isolation (PVI) as treatment for atrial fibrillation (AF) is not effective in up to 60% of patients with persistent AF; AF drivers outside of the pulmonary veins can contribute to AF recurrences after PVI. In this study we explored the potential use of local conduction velocity (CV) during pacing as a marker of left atrial (LA) substrate complexity.Methods: LA activation times were recorded for 7 AF patients during coronary sinus (CS) pacing before PVI using a Pentaray catheter. Activation times were relative to the CS pacing spike. LA activation locations were triangularized to calculate CV: the local direction and speed of the activation wave front. CV was quantified by the total CV distribution.Results: A mean of 1622 CVs were calculated per patient. Distribution of CVs showed a similar morphology, with median CVs in the range [0.26, 0.36] and interquartile ranges in the range [0.29, 0.39].Conclusion: This study shows that although it is feasible to calculate CVs based on sequential CARTO mapping of the LA during CS pacing, the resulting distribution of CVs using this procedure is not necessarily able to identify substrate complexity because of the large similarity between distributions and the relatively small differences in medians.","PeriodicalId":6697,"journal":{"name":"2019 Computing in Cardiology (CinC)","volume":"38 1","pages":"Page 1-Page 4"},"PeriodicalIF":0.0,"publicationDate":"2019-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"89655333","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 : 2019-09-01DOI: 10.23919/CinC49843.2019.9005767
J. V. Zaen, Elsa Genzoni, F. Braun, P. Renevey, E. Pruvot, J. Vesin, M. Lemay
Atrial fibrillation (AF) affects millions of individuals worldwide and can lead to serious complications such as stroke or heart failure. This arrhythmia is difficult to diagnose with ambulatory electrocardiogram monitors in the early stages due to its transient nature. Recent advances in wearable photoplethysmographic (PPG) devices are promising for screening AF in large populations as they are relatively comfortable and can be worn over long periods of time. Herein, we propose a system to detect AF from PPG recordings. This system is composed of a beat detector to extract interbeat intervals and a classifier for detection. We trained the classifier on a large public database of interbeat intervals and then evaluated the whole system on PPG recordings collected during catheter ablation procedures. We achieve an accuracy of 0.986 for the detection of AF with a sensitivity and specificity of 1.0 and 0.978 respectively. These metrics compare favorably with existing systems.
{"title":"Atrial Fibrillation Detection from PPG Interbeat Intervals via a Recurrent Neural Network","authors":"J. V. Zaen, Elsa Genzoni, F. Braun, P. Renevey, E. Pruvot, J. Vesin, M. Lemay","doi":"10.23919/CinC49843.2019.9005767","DOIUrl":"https://doi.org/10.23919/CinC49843.2019.9005767","url":null,"abstract":"Atrial fibrillation (AF) affects millions of individuals worldwide and can lead to serious complications such as stroke or heart failure. This arrhythmia is difficult to diagnose with ambulatory electrocardiogram monitors in the early stages due to its transient nature. Recent advances in wearable photoplethysmographic (PPG) devices are promising for screening AF in large populations as they are relatively comfortable and can be worn over long periods of time. Herein, we propose a system to detect AF from PPG recordings. This system is composed of a beat detector to extract interbeat intervals and a classifier for detection. We trained the classifier on a large public database of interbeat intervals and then evaluated the whole system on PPG recordings collected during catheter ablation procedures. We achieve an accuracy of 0.986 for the detection of AF with a sensitivity and specificity of 1.0 and 0.978 respectively. These metrics compare favorably with existing systems.","PeriodicalId":6697,"journal":{"name":"2019 Computing in Cardiology (CinC)","volume":"101 1","pages":"Page 1-Page 4"},"PeriodicalIF":0.0,"publicationDate":"2019-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"86242378","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 : 2019-09-01DOI: 10.23919/CinC49843.2019.9005754
ByeongTak Lee, Kyung-Jae Cho, Oyeon Kwon, Yeha Lee
Early prediction of sepsis is a clinically important, yet remains challenging. As machine learning develops, there have been many approaches for prediction of sepsis using neural network-based models. In this work, We propose various methods including feature engineering, regularization technique, and train data sampling methods, which can boost the performance of the model. Our approach consist of three-component: a feature engineering, an auxiliary loss, and a manipulation of training distribution. In feature engineering, we employed a novel input imputation method that combines input decay, masking, and duration of missing and input transformation. As for regularization, we used the reconstruction error as the auxiliary loss. Meanwhile, we manipulated the distribution of training sample using normal point re-sampling and population-based sampling. On the validation set, our approach improved the performance of LSTM as AUROC/AUPRC of 0. 045/0.017, and the performance of transformer is enhanced AUROC/AUPRC of 0.034/0.024. Finally, we submitted our transformer trained with proposed method on the official test set and obtained the utility score of 0.291 (Team name:vn, Rank:23).
{"title":"Improving the Performance of a Neural Network for Early Prediction of Sepsis","authors":"ByeongTak Lee, Kyung-Jae Cho, Oyeon Kwon, Yeha Lee","doi":"10.23919/CinC49843.2019.9005754","DOIUrl":"https://doi.org/10.23919/CinC49843.2019.9005754","url":null,"abstract":"Early prediction of sepsis is a clinically important, yet remains challenging. As machine learning develops, there have been many approaches for prediction of sepsis using neural network-based models. In this work, We propose various methods including feature engineering, regularization technique, and train data sampling methods, which can boost the performance of the model. Our approach consist of three-component: a feature engineering, an auxiliary loss, and a manipulation of training distribution. In feature engineering, we employed a novel input imputation method that combines input decay, masking, and duration of missing and input transformation. As for regularization, we used the reconstruction error as the auxiliary loss. Meanwhile, we manipulated the distribution of training sample using normal point re-sampling and population-based sampling. On the validation set, our approach improved the performance of LSTM as AUROC/AUPRC of 0. 045/0.017, and the performance of transformer is enhanced AUROC/AUPRC of 0.034/0.024. Finally, we submitted our transformer trained with proposed method on the official test set and obtained the utility score of 0.291 (Team name:vn, Rank:23).","PeriodicalId":6697,"journal":{"name":"2019 Computing in Cardiology (CinC)","volume":"22 1","pages":"Page 1-Page 4"},"PeriodicalIF":0.0,"publicationDate":"2019-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"87442264","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 : 2019-09-01DOI: 10.23919/CinC49843.2019.9005708
Thion Ming Chieng, Y. Hau, Z. Omar, Chiao Wen Lim
An efficient electrocardiogram (ECG) delineation algorithm is proposed to instantaneously delineate the ECG characteristic points, such as peak, onset and offset points of QRS, P and T waves. It is essential to delineate the ECG characteristic waves accurately and precisely as it ensure the performance of ECG analysis and diagnosis. The proposed delineation algorithm is based on discrete wavelet transform (DWT) and moving window average (MWA) techniques. The proposed delineation algorithm is evaluated and assessed with the annotation data of QT database in term of accuracy, sensitivity and positive predictive value. With the only available 13 sets QT database records with modified Lead II data, the proposed algorithm achieved significant P peak, R peak, T peak and T offset delineation performance with the accuracy of 95.34%, 99.80%, 90.82% and 86.33% respectively when evaluated with q1c annotation file. The mean difference between detected and annotated T offset based on q1c and q2c is 13 ms and 3.6 ms respectively. The delineation of 15 minute-long ECG record only required 74.702 second. As conclusion, the proposed ECG delineation algorithm based on DWT and MWA techniques have been proven simple, efficient and accurate in delineating the significant ECG characteristic points.
{"title":"An Efficient Instantaneous ECG Delineation Algorithm","authors":"Thion Ming Chieng, Y. Hau, Z. Omar, Chiao Wen Lim","doi":"10.23919/CinC49843.2019.9005708","DOIUrl":"https://doi.org/10.23919/CinC49843.2019.9005708","url":null,"abstract":"An efficient electrocardiogram (ECG) delineation algorithm is proposed to instantaneously delineate the ECG characteristic points, such as peak, onset and offset points of QRS, P and T waves. It is essential to delineate the ECG characteristic waves accurately and precisely as it ensure the performance of ECG analysis and diagnosis. The proposed delineation algorithm is based on discrete wavelet transform (DWT) and moving window average (MWA) techniques. The proposed delineation algorithm is evaluated and assessed with the annotation data of QT database in term of accuracy, sensitivity and positive predictive value. With the only available 13 sets QT database records with modified Lead II data, the proposed algorithm achieved significant P peak, R peak, T peak and T offset delineation performance with the accuracy of 95.34%, 99.80%, 90.82% and 86.33% respectively when evaluated with q1c annotation file. The mean difference between detected and annotated T offset based on q1c and q2c is 13 ms and 3.6 ms respectively. The delineation of 15 minute-long ECG record only required 74.702 second. As conclusion, the proposed ECG delineation algorithm based on DWT and MWA techniques have been proven simple, efficient and accurate in delineating the significant ECG characteristic points.","PeriodicalId":6697,"journal":{"name":"2019 Computing in Cardiology (CinC)","volume":"46 1","pages":"Page 1-Page 4"},"PeriodicalIF":0.0,"publicationDate":"2019-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"80009459","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 : 2019-09-01DOI: 10.23919/CinC49843.2019.9005824
G. Jiménez-Pérez, A. Alcaine, O. Camara
Automatic detection and delineation of the electrocardiogram (ECG) is usually the first step for many feature extraction tasks. Although deep learning (DL) approaches have been proposed in the literature, those employ non-optimal and outdated architectures. Thus, rule-based algorithms remain as state-of-the-art. Nevertheless, those may not generalize on other datasets and require difficult offline tuning for adapting to new scenarios. This work frames this task as a segmentation problem for using an adaptation of the U-Net architecture, a fully convolutional network. The detection performance shows a precision of 89.27%, 98.18% and 93.60% for the detection of the P, QRS and T waves, respectively, and a recall of 89.07%, 99.47% and 95.21%. This work shows promising results, outperforming existing DL approaches while being more generalizable than rule-based methods.
{"title":"U-Net Architecture for the Automatic Detection and Delineation of the Electrocardiogram","authors":"G. Jiménez-Pérez, A. Alcaine, O. Camara","doi":"10.23919/CinC49843.2019.9005824","DOIUrl":"https://doi.org/10.23919/CinC49843.2019.9005824","url":null,"abstract":"Automatic detection and delineation of the electrocardiogram (ECG) is usually the first step for many feature extraction tasks. Although deep learning (DL) approaches have been proposed in the literature, those employ non-optimal and outdated architectures. Thus, rule-based algorithms remain as state-of-the-art. Nevertheless, those may not generalize on other datasets and require difficult offline tuning for adapting to new scenarios. This work frames this task as a segmentation problem for using an adaptation of the U-Net architecture, a fully convolutional network. The detection performance shows a precision of 89.27%, 98.18% and 93.60% for the detection of the P, QRS and T waves, respectively, and a recall of 89.07%, 99.47% and 95.21%. This work shows promising results, outperforming existing DL approaches while being more generalizable than rule-based methods.","PeriodicalId":6697,"journal":{"name":"2019 Computing in Cardiology (CinC)","volume":"5 1","pages":"Page 1-Page 4"},"PeriodicalIF":0.0,"publicationDate":"2019-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"76390345","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：