Accurate non-invasive diagnoses in the context of cardiac diseases are problems that hitherto remain unresolved. We propose an unsupervised classification of atrial flutter (AFL) using dimensional transforms of ECG signals in high dimensional vector spaces. A mathematical model is used to generate synthetic signals based on clinical AFL signals, and hierarchical clustering analysis and novel machine learning (ML) methods are designed for the un-supervised classification. Metrics and accuracy parameters are created to assess the performance of the model, proving the power of this novel approach for the diagnosis of AFL from ECG using innovative AI algorithms.
{"title":"Classification of Atrial Tachycardia Types Using Dimensional Transforms of ECG Signals and Machine Learning","authors":"S. Ruipérez-Campillo, J. Millet-Roig, F. Castells","doi":"10.22489/CinC.2022.349","DOIUrl":"https://doi.org/10.22489/CinC.2022.349","url":null,"abstract":"Accurate non-invasive diagnoses in the context of cardiac diseases are problems that hitherto remain unresolved. We propose an unsupervised classification of atrial flutter (AFL) using dimensional transforms of ECG signals in high dimensional vector spaces. A mathematical model is used to generate synthetic signals based on clinical AFL signals, and hierarchical clustering analysis and novel machine learning (ML) methods are designed for the un-supervised classification. Metrics and accuracy parameters are created to assess the performance of the model, proving the power of this novel approach for the diagnosis of AFL from ECG using innovative AI algorithms.","PeriodicalId":117840,"journal":{"name":"2022 Computing in Cardiology (CinC)","volume":"133 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-09-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"123138374","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}
C. Antón, Jorge Sánchez, Andreas Heinkele, L. Unger, A. Haas, K. Schmidt, A. Luik, A. Loewe, O. Doessel
Regions with pathologically altered substrate have been identified as potential drivers for atrial fibrillation (AF) maintenance. Recently, local impedance (LI) measurements have gained attention as surrogate for atrial substrate assessment as it does not rely on electrical activity of the heart. However, an appropriate electrode-tissue contact force (CF) is needed and its effect on the LI measurements has not yet been characterized in depth. In this study, we applied several CF to a catheter in contact with a tissue patch modeled as healthy and scar atrial myocardium whose thickness was varied in anatomical ranges to study the impact of the mechanical deformation the LI measurements. When applying CF between 0 and 6g, in silico LI ranged from 160 $Omega$ to 175 $Omega$ in healthy my-ocardium, whereas 148 $Omega$ and 151 $Omega$ for scar tissue. Increasing CF in scar tissue up to 25 g, increased LI up to 156 $Omega$. The model was validated against clinically measured LI at different CF from AF patients. Simulation results applying identical CF in both tissues yielded lower LI values in scar. Moreover, LI increased in healthy and scar tissue when the thickness and CF were increased. Given the results of our study, we conclude that in silico experiments can not only distinguish between healthy and scar tissue by combining CF and LI, but also that our simulation environment represents clinical LI measurements with and without mechanical deformation in a tissue model.
{"title":"Effect of Contact Force on Local Electrical Impedance in Atrial Tissue - an In Silico Evaluation","authors":"C. Antón, Jorge Sánchez, Andreas Heinkele, L. Unger, A. Haas, K. Schmidt, A. Luik, A. Loewe, O. Doessel","doi":"10.22489/CinC.2022.337","DOIUrl":"https://doi.org/10.22489/CinC.2022.337","url":null,"abstract":"Regions with pathologically altered substrate have been identified as potential drivers for atrial fibrillation (AF) maintenance. Recently, local impedance (LI) measurements have gained attention as surrogate for atrial substrate assessment as it does not rely on electrical activity of the heart. However, an appropriate electrode-tissue contact force (CF) is needed and its effect on the LI measurements has not yet been characterized in depth. In this study, we applied several CF to a catheter in contact with a tissue patch modeled as healthy and scar atrial myocardium whose thickness was varied in anatomical ranges to study the impact of the mechanical deformation the LI measurements. When applying CF between 0 and 6g, in silico LI ranged from 160 $Omega$ to 175 $Omega$ in healthy my-ocardium, whereas 148 $Omega$ and 151 $Omega$ for scar tissue. Increasing CF in scar tissue up to 25 g, increased LI up to 156 $Omega$. The model was validated against clinically measured LI at different CF from AF patients. Simulation results applying identical CF in both tissues yielded lower LI values in scar. Moreover, LI increased in healthy and scar tissue when the thickness and CF were increased. Given the results of our study, we conclude that in silico experiments can not only distinguish between healthy and scar tissue by combining CF and LI, but also that our simulation environment represents clinical LI measurements with and without mechanical deformation in a tissue model.","PeriodicalId":117840,"journal":{"name":"2022 Computing in Cardiology (CinC)","volume":"336 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-09-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"116653246","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}
Rosie Barrows, M. Strocchi, Christoph M. Augustin, M. Gsell, C. Roney, J. Solís-Lemus, Hao Xu, K. Gillette, R. Rajani, J. Whitaker, E. Vigmond, M. Bishop, G. Plank, S. Niederer
Heart rate (HR) and effective atrial contraction affect left ventricular (LV) output. This is particularly relevant in atrial fibrillation (AF) patients, where HR is fast and irregular and atrial contraction almost completely absent. The effect of AF on the LV remains understudied, although a better understanding of these mechanisms could improve AF patient care. We have used a four-chamber electromechanics model to quantify how AF impacts LV function. Our model accounts for the effect of the pericardium and the coupling with the circulatory system, represented as a closed loop, providing physiological preload and afterload for the heart. The heart model was used for a factorial study with two HRs (70 bpm and 120 bpm) and in the presence and in the absence of atrial contraction. We found that an increased HR and lack of atrial contraction alone led to a small decrease in ejection fraction (42% to 40% and 42% to 41%, respectively). However, the interaction between an increased HR and lack of atrial contraction led to a drop in ejection fraction from 42% to 36%. This study demonstrates that our four-chamber heart models can be used to investigate the effect of rapid HR and ineffective atrial contraction on LV output and that AF can significantly impact LV function. This motivates further studies investigating the effect of AF on the whole heart.
{"title":"The Effect of Heart Rate and Atrial Contraction on Left Ventricular Function","authors":"Rosie Barrows, M. Strocchi, Christoph M. Augustin, M. Gsell, C. Roney, J. Solís-Lemus, Hao Xu, K. Gillette, R. Rajani, J. Whitaker, E. Vigmond, M. Bishop, G. Plank, S. Niederer","doi":"10.22489/CinC.2022.212","DOIUrl":"https://doi.org/10.22489/CinC.2022.212","url":null,"abstract":"Heart rate (HR) and effective atrial contraction affect left ventricular (LV) output. This is particularly relevant in atrial fibrillation (AF) patients, where HR is fast and irregular and atrial contraction almost completely absent. The effect of AF on the LV remains understudied, although a better understanding of these mechanisms could improve AF patient care. We have used a four-chamber electromechanics model to quantify how AF impacts LV function. Our model accounts for the effect of the pericardium and the coupling with the circulatory system, represented as a closed loop, providing physiological preload and afterload for the heart. The heart model was used for a factorial study with two HRs (70 bpm and 120 bpm) and in the presence and in the absence of atrial contraction. We found that an increased HR and lack of atrial contraction alone led to a small decrease in ejection fraction (42% to 40% and 42% to 41%, respectively). However, the interaction between an increased HR and lack of atrial contraction led to a drop in ejection fraction from 42% to 36%. This study demonstrates that our four-chamber heart models can be used to investigate the effect of rapid HR and ineffective atrial contraction on LV output and that AF can significantly impact LV function. This motivates further studies investigating the effect of AF on the whole heart.","PeriodicalId":117840,"journal":{"name":"2022 Computing in Cardiology (CinC)","volume":"26 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-09-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"116780075","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}
R. Abad, E. Franco, S. Ruipérez-Campillo, C. Lozano, F. Castells, J. Moreno
Persistent atrial fibrillation ablation has a high recurrence rate. In this work, we performed an analysis of bipolar intracavitary signals obtained with a conventional 24-pole diagnostic catheter (Woven Orbiter) placed in the right atrium and coronary sinus in a cohort of patients with persistent atrial fibrillation undergoing ablation to detect features predictive of acute procedural success (conversion to sinus rhythm during ablation) and the occurrence of recurrences. The goal is to arrive at a quantitative description of the degree of randomness of the atrial response in atrial fibrillation and to demonstrate the presence of hidden periodic components. This was done by the determination of the autocorrelation function. Results showed that higher correlation in relative maximum peaks, and a lower dominant atrial frequency (greater distance between relative amplitude maxima) may be associated with a greater likelihood of achieving reversion to sinus rhythm and lower probability of recurrences. A larger study is needed to draw conclusions.
{"title":"Autocorrelation Function for Predicting Arrhythmic Recurrences in Patients Undergoing Persistent Atrial Fibrillation Ablation","authors":"R. Abad, E. Franco, S. Ruipérez-Campillo, C. Lozano, F. Castells, J. Moreno","doi":"10.22489/CinC.2022.424","DOIUrl":"https://doi.org/10.22489/CinC.2022.424","url":null,"abstract":"Persistent atrial fibrillation ablation has a high recurrence rate. In this work, we performed an analysis of bipolar intracavitary signals obtained with a conventional 24-pole diagnostic catheter (Woven Orbiter) placed in the right atrium and coronary sinus in a cohort of patients with persistent atrial fibrillation undergoing ablation to detect features predictive of acute procedural success (conversion to sinus rhythm during ablation) and the occurrence of recurrences. The goal is to arrive at a quantitative description of the degree of randomness of the atrial response in atrial fibrillation and to demonstrate the presence of hidden periodic components. This was done by the determination of the autocorrelation function. Results showed that higher correlation in relative maximum peaks, and a lower dominant atrial frequency (greater distance between relative amplitude maxima) may be associated with a greater likelihood of achieving reversion to sinus rhythm and lower probability of recurrences. A larger study is needed to draw conclusions.","PeriodicalId":117840,"journal":{"name":"2022 Computing in Cardiology (CinC)","volume":"15 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-09-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"116888989","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}
Leto L. Riebel, Z. Wang, H. Martinez-Navarro, C. Trovato, J. Biasetti, R. S. Oliveira, R. D. Santos, Blanca A Rodríguez
Delivery of human pluripotent stem cell-derived cardiomyocytes (hPSC-CMs) is a potential therapy to improve cardiac function after injury. However, hPSCCMs express immature electrophysiological and structural properties and may be pro-arrhythmic. Our goal is to identify key factors determining arrhythmic risk of hPSC-CM therapy in the infarcted human ventricles through modelling and simulation. We model three densities of hPSC-CMs covering 4%, 22%, and 39% of the infarct and border zone and induce re-entry through ectopic stimulation. We furthermore simulate the effect of different therapeutic agents on re-entry susceptibility. Due to the increased refractory period of the hPSC-CMs, the vulnerable window increases from 20ms in control, to 60ms in the low- and 80ms in the medium- and high-density scenarios. Our results highlight the density-dependent effect of hPSC-CM delivery on arrhythmic risk after myocardial infarction and show the effect of therapeutic strategies on this increased re-entry susceptibility.
{"title":"Modelling and Simulation Reveals Density-Dependent Re-Entry Risk in The Infarcted Ventricles After Stem Cell-Derived Cardiomyocyte Delivery","authors":"Leto L. Riebel, Z. Wang, H. Martinez-Navarro, C. Trovato, J. Biasetti, R. S. Oliveira, R. D. Santos, Blanca A Rodríguez","doi":"10.22489/CinC.2022.392","DOIUrl":"https://doi.org/10.22489/CinC.2022.392","url":null,"abstract":"Delivery of human pluripotent stem cell-derived cardiomyocytes (hPSC-CMs) is a potential therapy to improve cardiac function after injury. However, hPSCCMs express immature electrophysiological and structural properties and may be pro-arrhythmic. Our goal is to identify key factors determining arrhythmic risk of hPSC-CM therapy in the infarcted human ventricles through modelling and simulation. We model three densities of hPSC-CMs covering 4%, 22%, and 39% of the infarct and border zone and induce re-entry through ectopic stimulation. We furthermore simulate the effect of different therapeutic agents on re-entry susceptibility. Due to the increased refractory period of the hPSC-CMs, the vulnerable window increases from 20ms in control, to 60ms in the low- and 80ms in the medium- and high-density scenarios. Our results highlight the density-dependent effect of hPSC-CM delivery on arrhythmic risk after myocardial infarction and show the effect of therapeutic strategies on this increased re-entry susceptibility.","PeriodicalId":117840,"journal":{"name":"2022 Computing in Cardiology (CinC)","volume":"16 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-09-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"117343241","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}
Aim: The George B. Moody PhysioNet Challenge 2022 raised problems of heart murmur detection and related abnormal cardiac function identification from phonocardiograms (PCGs). This work describes the novel approaches developed by our team, Revenger, to solve these problems. Methods: PCGs were resampled to 1000 $Hz$, then filtered with a Butterworth band-pass filter of order 3, cut-off frequencies 25 - 400 $H{z}$, and z-score normalized. $We$ used the multi-task learning $(MTL)$ method via hard parameter sharing to train one neural network (NN) model for all the Challenge tasks. We performed neural architecture searching among a set of network backbones, including multi-branch convolutional neural networks (CNNs), SE-ResNets, TResNets, simplified $wav2vec2$, etc. Based on a stratified splitting of the subjects, 20% of the public data was left out as a validation set for model selection. The AdamW optimizer was adopted, along with the OneCycle scheduler, to optimize the model weights. Results: Our murmur detection classifier received a weighted accuracy score of 0.736 (ranked 14th out of 40 teams) and a Challenge cost score of 12944 (ranked 19th out of 39 teams) on the hidden validation set. Conclusion: We provided a practical solution to the problems of detecting heart murmurs and providing clinical diagnosis suggestions from PCGs.
目的:George B. Moody PhysioNet Challenge 2022提出了心脏杂音检测和相关心音图(pcg)异常心功能识别的问题。这项工作描述了我们的团队复仇者开发的解决这些问题的新方法。方法:将pcg重采样至1000 $Hz$，然后用3阶Butterworth带通滤波器滤波，截止频率为25 ~ 400 $H{z}$，并将z分数归一化。我们使用了多任务学习(MTL)方法，通过硬参数共享来训练一个神经网络(NN)模型，用于所有挑战任务。我们在一组网络骨干中进行神经结构搜索，包括多分支卷积神经网络(cnn)、SE-ResNets、TResNets、简化的$wav2vec2$等。基于对受试者的分层划分，20%的公共数据被遗漏作为模型选择的验证集。采用AdamW优化器和OneCycle调度器来优化模型权重。结果:我们的杂音检测分类器在隐藏验证集上的加权准确率得分为0.736(在40个团队中排名第14)，挑战成本得分为12944(在39个团队中排名第19)。结论:通过心电图对心脏杂音的检测提供了切实可行的方法，为临床诊断提供了建议。
{"title":"Searching for Effective Neural Network Architectures for Heart Murmur Detection from Phonocardiogram","authors":"Hao Wen, Ji-Su Kang","doi":"10.22489/CinC.2022.130","DOIUrl":"https://doi.org/10.22489/CinC.2022.130","url":null,"abstract":"Aim: The George B. Moody PhysioNet Challenge 2022 raised problems of heart murmur detection and related abnormal cardiac function identification from phonocardiograms (PCGs). This work describes the novel approaches developed by our team, Revenger, to solve these problems. Methods: PCGs were resampled to 1000 $Hz$, then filtered with a Butterworth band-pass filter of order 3, cut-off frequencies 25 - 400 $H{z}$, and z-score normalized. $We$ used the multi-task learning $(MTL)$ method via hard parameter sharing to train one neural network (NN) model for all the Challenge tasks. We performed neural architecture searching among a set of network backbones, including multi-branch convolutional neural networks (CNNs), SE-ResNets, TResNets, simplified $wav2vec2$, etc. Based on a stratified splitting of the subjects, 20% of the public data was left out as a validation set for model selection. The AdamW optimizer was adopted, along with the OneCycle scheduler, to optimize the model weights. Results: Our murmur detection classifier received a weighted accuracy score of 0.736 (ranked 14th out of 40 teams) and a Challenge cost score of 12944 (ranked 19th out of 39 teams) on the hidden validation set. Conclusion: We provided a practical solution to the problems of detecting heart murmurs and providing clinical diagnosis suggestions from PCGs.","PeriodicalId":117840,"journal":{"name":"2022 Computing in Cardiology (CinC)","volume":"75 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-09-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"127289174","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}
Franz Ehrlich, Johannes Bender, Hagen Malberg, Miriam Goldammer
Arousals during sleep give deep insights into the patho-physiology of sleep disorders and sleep quality. Detecting arousals is a time-consuming process manually per-formed by a trained expert. The required measurement is performed on an inpatient basis and is uncomfortable for the patient. As arousals relate to the autonomic nervous system, they also reflect in the electrocardiogram, which is therefore a promising alternative biosignal. In this study, we developed a deep learning model for automatic detection of sleep arousals from heart rate. We developed our algorithm using 5323 recordings from the Sleep Heart Health Study. 1003 of them were held-out as test data. We derived RR intervals from the ECG and interpolated them into a 4 Hz signal. Next, we developed a convolutional neural network (CNN) for end-to-end event detection. Model output is a continuous arousal probabil-ity with a frequency of 1 Hz. The optimization resulted in a twelve-layer CNN that achieved a Cohens kappa of 0.47, an area under the precision-recall curve of 0.54 on hold-out test data. This study demonstrates the ability of machine learning to detect arousals during sleep from heart rate. As our approach uses only the heart rate, it is potentially trans-ferable to other signals, e.g. the photoplethysmogram.
{"title":"Automatic Sleep Arousal Detection Using Heart Rate From a Single-Lead Electrocardiogram","authors":"Franz Ehrlich, Johannes Bender, Hagen Malberg, Miriam Goldammer","doi":"10.22489/CinC.2022.080","DOIUrl":"https://doi.org/10.22489/CinC.2022.080","url":null,"abstract":"Arousals during sleep give deep insights into the patho-physiology of sleep disorders and sleep quality. Detecting arousals is a time-consuming process manually per-formed by a trained expert. The required measurement is performed on an inpatient basis and is uncomfortable for the patient. As arousals relate to the autonomic nervous system, they also reflect in the electrocardiogram, which is therefore a promising alternative biosignal. In this study, we developed a deep learning model for automatic detection of sleep arousals from heart rate. We developed our algorithm using 5323 recordings from the Sleep Heart Health Study. 1003 of them were held-out as test data. We derived RR intervals from the ECG and interpolated them into a 4 Hz signal. Next, we developed a convolutional neural network (CNN) for end-to-end event detection. Model output is a continuous arousal probabil-ity with a frequency of 1 Hz. The optimization resulted in a twelve-layer CNN that achieved a Cohens kappa of 0.47, an area under the precision-recall curve of 0.54 on hold-out test data. This study demonstrates the ability of machine learning to detect arousals during sleep from heart rate. As our approach uses only the heart rate, it is potentially trans-ferable to other signals, e.g. the photoplethysmogram.","PeriodicalId":117840,"journal":{"name":"2022 Computing in Cardiology (CinC)","volume":"25 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-09-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"124918010","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}
ECGs are highly distorted by the MRI environment, making automated ECG analysis highly difficult. This study aimed at implementing a machine-learning (ML) based heartbeat classifier, using hand-crafted features, for the automatic detection of ventricular heartbeats during MRI. A model was trained on the MIT-BIH Arrhythmia Database and assessed on an in-house database of ECG acquired inside a 1.5T MRI (ECG-MRI). Features were extracted for each heartbeat from single-lead ECG signals including QRS morphological features based on Hermite functions, and RR interval-based features. A support vector machine was trained to classify normal (N) and ventricular ectopic beats (V‘). The classifier achieved F1 scores of 0.85 on the V' class on the validation fold on the MIT-BIH database, while it only achieved F1 scores of 0.15 on the ECG-MRI database. The proposed heartbeat classifier was developed on the MIT-BIH arrhythmia database using temporal features and QRS morphological features based on the assumption they would be less distorted by the MRI environment. However, even if performance on MIT-BIH were acceptable (although slightly lower than state-of-the-art approaches), results were poor on the ECG-MRI database. The results highlight the need for further developments by suppressing MRI-related artifacts, and by retraining on MRI specific datasets.
{"title":"Automated Detection of Ventricular Heartbeats from Electrocardiogram (ECG) Acquired During Magnetic Resonance Imaging (MRI)","authors":"Pierre G Aublin, J. Felblinger, J. Oster","doi":"10.22489/CinC.2022.192","DOIUrl":"https://doi.org/10.22489/CinC.2022.192","url":null,"abstract":"ECGs are highly distorted by the MRI environment, making automated ECG analysis highly difficult. This study aimed at implementing a machine-learning (ML) based heartbeat classifier, using hand-crafted features, for the automatic detection of ventricular heartbeats during MRI. A model was trained on the MIT-BIH Arrhythmia Database and assessed on an in-house database of ECG acquired inside a 1.5T MRI (ECG-MRI). Features were extracted for each heartbeat from single-lead ECG signals including QRS morphological features based on Hermite functions, and RR interval-based features. A support vector machine was trained to classify normal (N) and ventricular ectopic beats (V‘). The classifier achieved F1 scores of 0.85 on the V' class on the validation fold on the MIT-BIH database, while it only achieved F1 scores of 0.15 on the ECG-MRI database. The proposed heartbeat classifier was developed on the MIT-BIH arrhythmia database using temporal features and QRS morphological features based on the assumption they would be less distorted by the MRI environment. However, even if performance on MIT-BIH were acceptable (although slightly lower than state-of-the-art approaches), results were poor on the ECG-MRI database. The results highlight the need for further developments by suppressing MRI-related artifacts, and by retraining on MRI specific datasets.","PeriodicalId":117840,"journal":{"name":"2022 Computing in Cardiology (CinC)","volume":"29 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-09-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"126037047","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}
"Matheus Araujo, Dewen Zeng, João Palotti, Xinrong Hu, Yiyu Shi, L. Pyles, Q. Ni
Congenital heart disease (CHD) is a major cause of death for newborns, especially in low resources countries, due to limited access to heart specialists for timely diagnosis. As part of the George B. Moody PhysioNet Challenge 2022, we propose an automatic algorithm to detect CHD murmurs from digitally recorded heart sounds annotated by specialists. To train and validate our model, we use a dataset with 5282 heart sounds collected from 1568 children in the Paraiba state of Brazil recorded from multiple auscultation locations. Our team, named One_Heart_Health, used a two-stage strategy that combines embeddings from Mel spectrograms generated from audio segments and a final classifier that combine those embeddings to deliver the final classification per individual. On the official hidden test, we reached a weighted accuracy score of 0.729 (ranked 17th out of 40) and a challenge cost score of 13283 (ranked 23th out of 39). In our internal 5-fold cross-validation experiments, our approach reached a sensitivity of 0.76 ± 0.10 and a specificity of 0.85 ± 0.11. We have shown that a deep learning approach for murmur detection has the potential to mimic heart specialists to provide timely identification of CHD.
先天性心脏病(CHD)是新生儿死亡的主要原因，特别是在资源匮乏的国家，因为获得心脏病专家及时诊断的机会有限。作为George B. Moody PhysioNet Challenge 2022的一部分，我们提出了一种自动算法，从专家注释的数字记录的心音中检测冠心病杂音。为了训练和验证我们的模型，我们使用了一个包含5282个心音的数据集，这些心音来自巴西帕拉伊巴州的1568名儿童，记录于多个听诊位置。我们的团队名为One_Heart_Health，采用了两阶段策略，将音频片段生成的Mel谱图的嵌入和最终分类器结合起来，最终分类器将这些嵌入结合起来，为每个人提供最终分类。在官方的隐藏测试中，我们的加权准确率得分为0.729(在40个中排名第17)，挑战成本得分为13283(在39个中排名第23)。在我们的内部5倍交叉验证实验中，我们的方法达到了0.76±0.10的灵敏度和0.85±0.11的特异性。我们已经证明，一种用于杂音检测的深度学习方法有可能模仿心脏病专家，及时识别冠心病。
{"title":"Maiby's Algorithm: A Two-Stage Deep Learning Approach for Murmur Detection in Mel Spectrograms for Automatic Auscultation of Congenital Heart Disease","authors":"\"Matheus Araujo, Dewen Zeng, João Palotti, Xinrong Hu, Yiyu Shi, L. Pyles, Q. Ni","doi":"10.22489/CinC.2022.249","DOIUrl":"https://doi.org/10.22489/CinC.2022.249","url":null,"abstract":"Congenital heart disease (CHD) is a major cause of death for newborns, especially in low resources countries, due to limited access to heart specialists for timely diagnosis. As part of the George B. Moody PhysioNet Challenge 2022, we propose an automatic algorithm to detect CHD murmurs from digitally recorded heart sounds annotated by specialists. To train and validate our model, we use a dataset with 5282 heart sounds collected from 1568 children in the Paraiba state of Brazil recorded from multiple auscultation locations. Our team, named One_Heart_Health, used a two-stage strategy that combines embeddings from Mel spectrograms generated from audio segments and a final classifier that combine those embeddings to deliver the final classification per individual. On the official hidden test, we reached a weighted accuracy score of 0.729 (ranked 17th out of 40) and a challenge cost score of 13283 (ranked 23th out of 39). In our internal 5-fold cross-validation experiments, our approach reached a sensitivity of 0.76 ± 0.10 and a specificity of 0.85 ± 0.11. We have shown that a deep learning approach for murmur detection has the potential to mimic heart specialists to provide timely identification of CHD.","PeriodicalId":117840,"journal":{"name":"2022 Computing in Cardiology (CinC)","volume":"48 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-09-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"126131542","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}
Mohammad L. Karim, A. Bosnjak, J. Mclaughlin, P. Crawford, D. McEneaney, O. Escalona
This work focuses on the thermal analysis of a transdermal wireless radiofrequency (RF) energy transfer system, to power artificial heart pumps, particularly left-ventricular assist devices (LVADs). We aim to understand the blood perfusion factors to mitigate the skin heating effects and thermal injury to subcutaneous tissue under the RF coupling area. A 2-channel RF power loss emulator (RFPLE) system was developed to conduct a study independent of the wireless RF supply coupling method. The heating coils were implanted subcutaneously 6–8 mm beneath the porcine model skin. Heating effects due to RF coupling inefficiency power losses for conventional and our novel pulsed transmission waveform protocol were emulated. The thermal profiles were studied for varying levels of LVAD power requirement. An in-silico model was developed in parallel with the in-vivo experiments to aid the interpretation of results.
{"title":"Harnessing Dermal Blood Flow to Mitigate Skin Heating Effects in Wireless Transdermal Energy Systems for Driving Heart Pumps","authors":"Mohammad L. Karim, A. Bosnjak, J. Mclaughlin, P. Crawford, D. McEneaney, O. Escalona","doi":"10.22489/CinC.2022.409","DOIUrl":"https://doi.org/10.22489/CinC.2022.409","url":null,"abstract":"This work focuses on the thermal analysis of a transdermal wireless radiofrequency (RF) energy transfer system, to power artificial heart pumps, particularly left-ventricular assist devices (LVADs). We aim to understand the blood perfusion factors to mitigate the skin heating effects and thermal injury to subcutaneous tissue under the RF coupling area. A 2-channel RF power loss emulator (RFPLE) system was developed to conduct a study independent of the wireless RF supply coupling method. The heating coils were implanted subcutaneously 6–8 mm beneath the porcine model skin. Heating effects due to RF coupling inefficiency power losses for conventional and our novel pulsed transmission waveform protocol were emulated. The thermal profiles were studied for varying levels of LVAD power requirement. An in-silico model was developed in parallel with the in-vivo experiments to aid the interpretation of results.","PeriodicalId":117840,"journal":{"name":"2022 Computing in Cardiology (CinC)","volume":"34 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-09-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"126189591","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：