Gotowa bibliografia na temat „ELECTROCARDIOGRAM FEATURES”

This work is devoted to solving the scientific and practical problem of morphological analysis of electrocardiograms based on an integral biomedical signal with locally concentrated features. In modern conditions of introduction of telemedicine in the health care system of Ukraine the creation of cardiological decision support systems based on automatic morphological analysis of electrocardiogram is of particular importance. The authors proposed a method for synthesizing an integral electrocardiogram in the frontal plane from all limb leads, taking into account the lead angle in the hexaxial reference system and the position of the heart’s electrical axis, since integral electrocardiological signals allow to obtain more accurate results compared to conventional electrocardiogram, because they take into account the individual characteristics of patients, a wide variety of electrocardiogram waveforms and complexes, which is associated not only with the presence of pathological processes in the myocardium, but also with the position of the electrical axis of the heart, in particular, the electrocardiogram will not register a low-amplitude P wave in the II department in the case of a horizontal electrical axis, but it will be clearly visible on the integral signal. To implement the method proposed in the article, a program was written in the MATLAB language, , the high speed of computation and good optimization of which allow to obtain results much faster and more accurate than using traditional approaches, and using the MATLAB Runtime library, which does not require licensing and is distributed free of charge, it was possible to provide more economical development, as well as to implement interaction with popular operating systems, which makes it more accessible and versatile. Verification of the results was carried out using a database of electrocardiograms, which were recorded using a transtelephone digital 12-channel electrocardiological complex “Telecard”, which is part of the medical diagnostic complex “TREDEX”. The paper shows that the proposed method for the synthesis of an integral signal with locally concentrated features will improve the quality of morphological analysis of electrocardiograms in cardiological decision support systems.

Cardiovascular disease is one of the main causes of death worldwide. Arrhythmias are an important group of cardiovascular diseases. The standard 12-lead electrocardiogram signals are an important tool for diagnosing arrhythmias. Although 12-lead electrocardiogram signals provide more comprehensive arrhythmia information than single-lead electrocardiogram signals, it is difficult to effectively fuse information between different leads. In addition, most of the current researches working on automatic diagnosis of cardiac arrhythmias are based on modeling and analysis of single-mode features extracted from one-dimensional electrocardiogram sequences, ignoring the frequency domain features of electrocardiogram signals. Therefore, developing an automatic arrhythmia detection algorithm based on 12-lead electrocardiogram with high accuracy and strong generalization ability is still challenging. In this paper, a multimodal feature fusion model based on the mechanism is developed. This model utilizes a dual channel deep neural network to extract different dimensional features from one-dimensional and two-dimensional electrocardiogram time–frequency maps, and combines attention mechanism to effectively fuse the important features of 12-lead, thereby obtaining richer arrhythmia information and ultimately achieving accurate classification of nine types of arrhythmia signals. This study used electrocardiogram signals from a mixed dataset to train, validate, and evaluate the model, with an average of F1 score and average accuracy reached 0.85 and 0.97, respectively. Experimental results show that our algorithm has stable and reliable performance, so it is expected to have good practical application potential.

The authors of this paper present a new method to characterize the electrocardiogram (ECG) for individual identification. We propose an ECG biometric system which is insensitive to noise signals and muscle flexure. The method utilizes the principal of linearly projecting the heartbeat features into a subspace of lower dimension using an orthogonal basis that represents the most significant features to distinguish the individuals. The performance of the proposed biometric system is evaluated on the subjects of both health statuses such as the ECG recordings of MIT-BIH Arrhythmia database and the ECG recordings of normal subjects prepared at IIT(BHU). The result demonstrates that the derived eigenbeat features from proposed ECG characterization perform better and achieve the recognition accuracy of 91.42% and 95.55% on the subjects of MIT-BIH Arrhythmia database and IIT(BHU) database, respectively.

PurposeDigital computing and machine learning-driven predictive analysis in the diagnosis of non-communicable diseases are gaining significance. Globally many research studies are focusing on developing comprehensive models for such detection. Categorically in the proposed diagnosis for arrhythmia, which is a critical diagnosis to prevent cardiac-related deaths, any constructive models can be a value proposition. In this study, the focus is on developing a holistic system that predicts the scope of arrhythmia from the given electrocardiogram report. The proposed method is using the sequential patterns of the electrocardiogram elements as features.Design/methodology/approachConsidering the decision accuracy of the contemporary classification methods, which is not adequate to use in clinical practices, this manuscript coined a new dimension of features to perform supervised learning and classification using the AdaBoost classifier. The proposed method has titled “Electrocardiogram stream level correlated patterns as features (ESCPFs),” which takes electrocardiograms (ECGs) signal streams as input records to perform supervised learning-based classification to detect the arrhythmia scope in given ECG record.FindingsFrom the results and comparative reports generated for the study, it is evident that the model is performing with higher accuracy compared to some of the earlier models. However, focusing on the emerging solutions and technologies, if the accuracy factors for the model can be improved, it can lead to compelling predictions and accurate outcome from the process.Originality/valueThe authors represent complete automatic and rapid arrhythmia as classifier, which could be applied online and examine long ECG records sequence efficiently. By releasing the needs for extraction of features, the authors project an application based on raw signals, one result to heart rates date, whose objective is to lessen computation time when attaining minimum classification error outcomes.

In this paper, the conventional Stockwell transform is effectively used to classify the ECG arrhythmias. The performance of ECG classification mainly depends on feature extraction based on an efficient formation of morphological and temporal features and the design of the classifier. Feature extraction is the important component of designing the system based on pattern recognition since even the best classifier will not perform better if the good features are not selected properly. Here, the S-transform (ST) is used to extract the morphological features which is appended with temporal features. This feature set is independently classified using artificial neural network (NN) and support vector machine (SVM). In this work, five classes of ECG beats (normal, ventricular, supra ventricular, fusion and unknown beats) from Massachusetts Institute of Technology-Beth Israel Hospital (MIT-BIH) arrhythmia database are classified according to AAMI EC57 1998 standard (Association for the Advancement of Medical Instrumentation). Performance is evaluated on several normal and abnormal ECG signals of MIT-BIH arrhythmias database using two classifier techniques: ST with NN classifier (ST-NN) and other proposed ST with SVM classifier (ST-SVM). The proposed method achieves accuracy of 98.47%. The performance of the proposed technique is compared with ST-NN and earlier reported technique.

Most existing electrocardiogram (ECG) feature extraction methods rely on rule-based approaches. It is difficult to manually define all ECG features. We propose an unsupervised feature learning method using a convolutional variational autoencoder (CVAE) that can extract ECG features with unlabeled data. We used 596,000 ECG samples from 1,278 patients archived in biosignal databases from intensive care units to train the CVAE. Three external datasets were used for feature validation using two approaches. First, we explored the features without an additional training process. Clustering, latent space exploration, and anomaly detection were conducted. We confirmed that CVAE features reflected the various types of ECG rhythms. Second, we applied CVAE features to new tasks as input data and CVAE weights to weight initialization for different models for transfer learning for the classification of 12 types of arrhythmias. The f1-score for arrhythmia classification with extreme gradient boosting was 0.86 using CVAE features only. The f1-score of the model in which weights were initialized with the CVAE encoder was 5% better than that obtained with random initialization. Unsupervised feature learning with CVAE can extract the characteristics of various types of ECGs and can be an alternative to the feature extraction method for ECGs.

Introduction Arrhythmogenic right ventricular cardiomyopathy (ARVC) is a genetic heart muscle disease characterized by electrical instability leading to ventricular arrhythmias and sudden cardiac death. The hallmark pathological lesion of ARVC is the transmural loss of the myocardium of the right ventricular (RV) free wall with replacement by fibro-fatty tissue. Three-dimensional electroanatomic voltage mapping (EVM) by CARTO system (Biosense-Webster, Diamond Bar, California) allows identification and characterization of low-voltage regions, i.e. "electroanatomical scars" (EAS), which in patients with ARVC correspond to areas of fibro-fatty replacement. Although the technique has been demonstrated to enhance the accuracy for diagnosing ARVC, its value for arrhythmic risk stratification remains to be established. Furthermore, the clinical utility of EVM for scar quantification and risk assessment is limited by its invasive nature, low availability and high costs. Thus, in daily clinical practice there is the need of a non-invasive test such as 12-lead electrocardiogram (ECG) for prediction of the amount of RV myocardial scar lesion and assessment of arrhythmic risk. Previous studies demonstrated an association between ECG repolarization/depolarization abnormalities and RV mechanical dilation/dysfunction. In fact, T wave inversion in right pericardial leads is the most common ECG abnormality of ARVC. However, the presence of T wave inversion in leads V1-V3, known as persistence of the juvenile pattern of repolarization, may also be observed in about 3% of healthy adults. The current perspective is that, at variance with healthy subjects, right precordial NTWs persist with exercise in ARVC patients. However, this view is not supported by systematic scientific data. Objective In this work, we aimed to further study some of the electrocardiographic features of ARVC. First, we assessed the prognostic value of EAS detected by EVM and its correlation with various non-invasive characteristics of ARVC, including abnormalities detected by surface ECG. Second, we studied the exercise-induced changes in right precordial negative T waves in patients with ARVC and in a group of healthy young individuals with persistence of the juvenile repolarization pattern Methods and results We first studied 69 consecutive ARVC patients (47 males; median age 35 years [28-45]) who underwent electrophysiological study and both bipolar and unipolar EVM. The extent of confluent bipolar (<1.5 mV) and unipolar (<6.0 mV) low-voltage electrograms was estimated using the CARTO-incorporated area calculation software. Fifty-three patients (77%) showed ≥1 RV electroanatomic scars with an estimated burden of bipolar versus unipolar low voltage areas of 24.8% (7.2-31.5) and 64.8% (39.8-95.3), respectively (P=0.009). In the remaining patients with normal bipolar EVM (n=16; 23%), the use of unipolar EVM unmasked ≥1 region of low-voltage electrogram affecting 26.2% (11.6-38.2) of RV wall. During a median follow-up of 41 (28-56) months, 19 (27.5%) patients experienced arrhythmic events. At multivariate analysis, the only independent predictor was the bipolar low-voltage electrogram burden (hazard ratio=1.6 per 5%; 95% confidence interval, 1.2-1.9; P<0.001). Patients with normal bipolar EVM had an uneventful clinical course. Then we further analyzed a subgroup including 49 patients [38 males, median age 35 years] with ARVC and an abnormal EVM by CARTO system. At univariate analysis, the presence of epsilon waves, the degree of RV dilation, the severity of RV dysfunction and the extent of negative T-waves correlated with RV-EAS% area. At multivariate analysis, the extent of negative T-waves remained the only independent predictor of RV-EAS% area (B=4.4, 95%CI 1.3-7.4, p=0.006) and correlated with the arrhythmic event-rate during follow-up (p=0.03). In a different cohort, we assessed the prevalence and relation to the clinical phenotype of exercise-induced right precordial negative T wave changes in 35 ARVC patients (19 males, mean age 22.2±6.2 years). Forty-one healthy individuals with right-precordial negative T waves served as controls. At peak of exercise, negative T waves persisted in 3 ARVC (9%) patients, completely normalized in 12 (34%) and partially reverted in 20 (57%). ARVC patients with or without negative T waves normalization showed a similar clinical phenotype. The overall prevalence of right precordial T-waves changes during exercise (normalization plus partial reversal) did not differ between ARVC patients and controls (92% versus 88%, p=1.0), while there was a statistically non significant trend towards a higher prevalence of complete normalization in controls (59% versus 34%, p=0.06). Conclusion In conclusion, our results showed that the extent of bipolar RV endocardial low-voltage area was a powerful predictor of arrhythmic outcome in ARVC independently of arrhythmic history and RV dilatation/dysfunction. A normal bipolar EVM characterized a low-risk subgroup of ARVC patients. Patients with abnormal ECG have a more severe RV EAS involvement, which is proportional to the extent of T wave inversion across ECG 12-leads and a higher arrhythmic risk. The absence of negative T waves characterizes a low-risk subgroup of ARVC patients with a more favorable clinical course because of a low rate of arrhythmic events. The results also showed that exercise-induced changes of negative T waves were unrelated to ARVC phenotypic manifestations and were of limited value for the differential diagnosis between ARVC and benign persistence of the juvenile repolarization pattern
Introduzione La cardiomiopatia aritmogena del ventricolo destro (CAVD) è una patologia genetica del muscolo cardiaco caratterizzata da instabilità elettrica che può portare a aritmie ventricolari e morte improvvisa. Dal punto di vista patologico, la CAVD si caratterizza per una progressiva perdita di tessuto miocardico della parete libera del ventricolo destro (VD) con sostituzione fibro-adiposa. Il mapaggio elettroanatomico tridimensionale (endocardial voltage mapping, EVM) col sistema CARTO (Biosense-Webster, Diamond Bar, California) consente di identificare e caratterizzare aree di basso-voltaggio, dette "cicatrici elettroanatomiche" (CEA), che in pazienti affetti da CAVD corrispondono ad aree di sostituzione fibro-adiposa. Nonostante la tecnica abbia dimostrato di migliorare l"accuratezza per la diagnosi di CAVD, il suo valore per la stratificazione del rischio aritmico rimane da dimostrate. Inoltre, l"utilità dell"EVM per la quantificazione della cicatrice e la valutazione del rischio è limitata dalla natura invasiva, bassa disponibilità ed alti costi. Quindi, nella pratica clinica quotidiana è auspicabile la disponibilità di un esame non-invasivo, come l"elettrocardiogramma (ECG), per la stima dell'estensione della CEA e la stratificazione del rischio aritmico. Studi precedenti hanno dimostrato un"associazione tra la presenza di anomalie della ripolarizzazione o della depolarizzazione all"ECG e l"entità della dilatazione e della disfunzione del VD. In particolare, l"inversione delle onde T nelle derivazioni precordiali destre V1-V3 è uno dei segni distintivi della CAVD. Tuttavia, lo stesso segno ECG può essere riscontrato come "persistenza del pattern giovanile di ripolarizzazione" fino al 3% degli adulti sani. La prospettiva attuale è che le T negative persistano con l'esercizio nei pazienti con CAVD ma non nei soggetti sani. Tuttavia, questa idea non è supportata da dati scientifici. Obbiettivo L"obbiettivo dell"attività di ricerca è stato quello di caratterizzare ulteriormente alcune delle caratteristiche ECG della CAVD. Inizialmente, abbiamo valutato il valore prognostico della CEA all"EVM e la sua correlazione con vari esami non invasivi, in particolare l"ECG. In secondo luogo, abbiamo studiato le modificazioni indotte dall"esercizio nella T negative nelle derivazioni precordiali destre in un gruppo di pazienti con CAVD ed in un gruppo di soggetti sani con "persistenza del pattern giovanile di ripolarizzazione". Metodi e risultati Sono stati studiati 69 pazienti consecutivi affetti da CAVD (47 maschi, età mediana 35 [28-45] anni) che sono stati sottoposti a studio elettrofisiologico endocavitario con mappa di voltaggio unipolare e bipolare. L"estensione delle aree contenenti elettrogrammi di basso voltaggio bipolari (<1.5 mV) e/o unipolari (<6.0 mV) è stata stimata usando un software incorporato nel sistema CARTO. Cinquantatre pazienti (77%) mostravano ≥1 CEA con un"estensione pari a 24.8% (7.2-31.5) dell"estensione del VD alla mappa bipolare e del 64.8% (39.8-95.3) alla mappa unipolare (p=0.009). Nei rimanenti 16 pazienti con mappa bipolare normale, la mappa unipolare è risultata alterata con un"estensione delle lesioni pari al 26.2% (11.6-38.2) del VD. Nel corso di un follow-up medio di 41 (28-56) mesi, 19 (27.5%) pazienti hanno avuto un evento aritmico maggiore. All'analisi multivariata, l'unico predittore indipendente di eventi aritmici è risultata l'estensione della CEA alla mappa di voltaggio bipolare (hazard ratio=1.6 per 5%; intervallo di confidenza 95%: 1.2-1.9; P<0.001). I pazienti con mappa di voltaggio bipolare negativa hanno avuto un follow-up privo di eventi. Successivamente, abbiamo analizzato un sottogruppo di 49 pazienti (38 maschi, età mediana 35 anni) con CAVD e mappa di voltaggio bipolare positiva. All'analisi univariata, la presenza di onde epsilon, il grado di dilatazione del VD, la severità della disfunzione del VD e l'estensione delle T negative all'ECG correlavano con l'estensione della CEA alla mappa bipolare. All'analisi multivariata, l'estensione delle onde T negative è rimasta l'unico predittore di estensione della CEA (B=4.4, 95%CI 1.3-7.4, p=0.006). Questo parametro si è inoltre dimostrato correlare con il rischio di eventi aritmici durante il follow-up (p=0.03). In una coorte differente, abbiamo valutato il comportamento durante test da sforzo delle T negative nelle derivazioni precordiali destre V1-V4 in 35 pazienti con CAVD (19 maschi, età media 22.2"±6.2 anni) ed in 41 controlli appaiati per età e sesso con benigna "persistenza del pattern giovanile di ripolarizzazione". Al picco dell'esercizio, le onde T negative persistevano in 3 (9%) pazienti con CAVD, normalizzavano completamente in 12 (34%) e normalizzavano parzialmente in 20 (57%). I pazienti affetti da CAVD con e senza normalizzazione delle onde T durante l'esercizio mostravano un fenotipo simile. La prevalenza di normalizzazione (parziale o completa) delle onde T era simile nei pazienti e nei controlli (92% e 88%, p=1,0), mentre si è notato un trend non significativo verso una più alta prevalenza di normalizzazione completa nei controlli sani rispetto ai pazienti con CAVD (59% e 34%, p=0,06). Conclusioni In conclusione, i nostri risultati hanno mostrato che l'estensione della CEA bipolare alla mappa di voltaggio del VD è un potente predittore di rischio aritmico nei pazienti con CAVD, indipendentemente dalla storia aritmica e dal grado di disfunzione/dilatazione del VD. Una mappa di voltaggio bipolare normale caratterizza una popolazione di pazienti con CAVD a basso rischio. Abbiamo inoltre dimostrato che l'estensione della CEA bipolare può essere stimata dall'estensione delle anomalie della ripolarizzazione (T negative) all'ECG. I pazienti che non mostrano T negative all'ECG dimostrano un basso rischio aritmico. Infine, abbiamo dimostrato che il comportamento delle T negative nelle derivazioni precordiali destre V1-V3 non è un utile strumento di diagnosi differenziale tra CAVD e benigna "persistenza del pattern giovanile di ripolarizzazione"

Atrial fibrillation is a common heart arrhythmia which is characterized by a missing or irregular contraction of the atria. The disease is a risk factor for other more serious diseases and the total medical costs in society are extensive. Therefore it would be beneficial to improve and optimize the prevention and detection of the disease.   Pulse palpation and heart auscultation can facilitate the detection of atrial fibrillation clinically, but the diagnosis is generally confirmed by an ECG examination. Today there are several algorithms that detect atrial fibrillation by analysing an ECG. A common method is to study the heart rate variability (HRV) and by different types of statistical calculations find episodes of atrial fibrillation which deviates from normal sinus rhythm.   Two algorithms for detection of atrial fibrillation have been evaluated in Matlab. One is based on the coefficient of variation and the other uses a logistic regression model. Training and testing of the algorithms were done with data from the Physionet MIT database. Several steps of signal processing were used to remove different types of noise and artefacts before the data could be used.   When testing the algorithms, the CV algorithm performed with a sensitivity of 91,38%, a specificity of 93,93% and accuracy of 92,92%, and the results of the logistic regression algorithm was a sensitivity of 97,23%, specificity of 93,79% and accuracy of 95,39%. The logistic regression algorithm performed better and was chosen for implementation in Java, where it achieved a sensitivity of 97,31%, specificity of 93,47% and accuracy of 95,25%.
Förmaksflimmer är en vanlig hjärtrytmrubbning som kännetecknas av en avsaknad eller oregelbunden kontraktion av förmaken. Sjukdomen är en riskfaktor för andra allvarligare sjukdomar och de totala kostnaderna för samhället är betydande. Det skulle därför vara fördelaktigt att effektivisera och förbättra prevention samt diagnostisering av förmaksflimmer.   Kliniskt diagnostiseras förmaksflimmer med hjälp av till exempel pulspalpation och auskultation av hjärtat, men diagnosen brukar fastställas med en EKG-undersökning. Det finns idag flertalet algoritmer för att detektera arytmin genom att analysera ett EKG. En av de vanligaste metoderna är att undersöka variabiliteten av hjärtrytmen (HRV) och utföra olika sorters statistiska beräkningar som kan upptäcka episoder av förmaksflimmer som avviker från en normal sinusrytm.   I detta projekt har två metoder för att detektera förmaksflimmer utvärderats i Matlab, en baseras på beräkningar av variationskoefficienten och den andra använder sig av logistisk regression. EKG som kommer från databasen Physionet MIT används för att träna och testa modeller av algoritmerna. Innan EKG-signalen kan användas måste den behandlas för att ta bort olika typer av brus och artefakter.   Vid test av algoritmen med variationskoefficienten blev resultatet en sensitivitet på 91,38%, en specificitet på 93,93% och en noggrannhet på 92,92%. För logistisk regression blev sensitiviteten 97,23%, specificiteten 93,79% och noggrannheten 95,39%. Algoritmen med logistisk regression presterade bättre och valdes därför för att implementeras i Java, där uppnåddes en sensitivitet på 91,31%, en specificitet på 93,47% och en noggrannhet på 95,25%.

Dissertação de Mestrado Integrado em Engenharia Electrotécnica e de Computadores apresentada à Faculdade de Ciências e Tecnologia
Recentemente, devido ao aumento do número de mortes por doenças cardiovasculares, o diagnóstico de doenças cardíacas tem sido um foco de bastante interesse no mundo computacional. Isto porque, a deteção de doenças cardíacas num estágio inicial pode prolongar a vida através de um tratamento adequado.Inúmeros métodos para fazer a monitorização das condições cardíacas foram introduzidos no mercado, sendo que o mais utilizado é a eletrocardiograma (ECG). O ECG é o registo da variação da atividade bioelétrica do coração, que representa as contrações e relaxamentos cíclicos do músculo cardíaco humano. Este fornece informações importantes sobre os aspetos funcionais do coração e do sistema cardiovascular. No entanto, ler grandes quantidades de sinais de ECGs é um processo demorado. Por isso, a deteção automática de anomalias nos sinais do eletrocardiograma atua como um assistente para os médicos diagnosticarem uma condição cardíaca.As irregularidades presentes no batimento cardíaco no formato do ECG são geralmente chamada de arritmia. Arritmia é um termo comum para qualquer distúrbio cardíaco que difere do ritmo normal. A análise automática do sinal de ECG para deteção de batimentos cardíacos é difícil devido à grande variação nas características morfológicas e temporais das formas de onda do ECG entre pacientes diferentes, bem como nos mesmos pacientes.Esta dissertação tem como objetivo desenvolver um método de deteção da fibrilhação auricular, que é um tipo de arritmia, através do sinal do eletrocardiograma. Deste modo, a metodologia proposta baseia-se na extração de um conjunto de características ("features'') dos ECG, e na sua classificação através de diferentes tipos de classificadores baseados em "machine learning", que consiste na execução de algoritmos que criam de modo automático modelos com base num conjunto de dados.O método desenvolvido foi utilizado em 2000 ECG de maneira a determinar a eficácia de cada um dos classificadores na deteção da doença cardíaca. Deste modo o presente documento inclui um estudo sobre os parâmetros do modelo escolhido e os resultados de classificação correspondestes. Não só, apresenta uma análise sobre a influência de certos fatores no desempenho do sistema, nomeadamente o tamanho do conjunto de dados e o conjunto de "features" utilizado na representação.Por sua vez, os resultados obtidos apoiam o uso de classificadores baseados em "machine learning" como ferramenta de classificação, na área de deteção de doenças cardíacas. O sistema desenvolvido classifica 2000 ECG, provenientes de duas classes, normal e fibrilhação auricular, com uma taxa de eficácia global de 93%, para o conjunto de teste. Na deteção particular da fibrilhação auricular, registou-se uma eficácia de 97% para o conjunto de teste.
Recently, due to the increase in the number of deaths from cardiovascular diseases, the diagnosis of heart disease has been a focus of great interest in the computational world. This is because the detection of heart disease at an early stage can prolong life through proper treatment. Numerous methods for monitoring cardiac conditions have been introduced in the market, the most used being the electrocardiogram (ECG). The ECG is the recording of the variation in the bioelectric activity of the heart, which represents the contractions and cyclical relaxations of the human cardiac muscle. It provides important information about the functional aspects of the heart and the cardiovascular system. However, processing large amounts of raw electrocardiogram signals from sensors is time consuming. Therefore, the automatic detection of abnormalities in the electrocardiogram signals acts as an assistant for doctors to diagnose a heart condition.The irregularities present in the heartbeat in the ECG format are generally called arrhythmia. Arrhythmia is a common term for any heart disorder that differs from the normal rhythm. The automatic analysis of the ECG signal to detect heartbeat is difficult due to the great variation in the morphological and temporal characteristics of the ECG waveforms between different patients, as well as in the same patients. This dissertation aims to develop a method of detecting atrial fibrillation, which is a type of arrhythmia, using the electrocardiogram signal. The proposed methodology is based on the extraction of a set of characteristics (features) from ECG, and on their classification through different types of classifiers based on machine learning. These classifiers consists of the execution of algorithms that automatically create representation models based on a set of data.The method developed was used on 2000 ECG in order to determine the effectiveness of each of the classifiers in detecting heart disease. Thus, this document includes a study on the parameters of the chosen model and the corresponding classification results. Not only, it presents an analysis of the influence of certain factors on the performance of the system, namely the size of the data set and the set of features used in the representation.In turn, the results obtained support the use of classifiers based on machine learning as a classification tool, in the area of heart disease detection. Therefore, the developed system classifies 2000 ECG, as two classes: normal and atrial fibrillation, with an overall effectiveness rate of 93%, on the test set. In the particular detection of atrial fibrillation, there was an efficiency of 97% for the test set.

碩士
國立清華大學
電機工程學系
102
Nowadays, sleep disorders become an important issue because it can adversely affect neurocognitive, cardiovascular, respiratory diseases, which subsequently induce the behavior disorder, majority of these cases up to 85% of these cases are obstructive sleep apnea (OSA). Therefore, the study of how to diagnose, detect and treat OSA is becoming a critical issue from both academical and medical perspective. Polysomnography (PSG) can monitor the OSA using relatively few invasive techniques. However, sleep studies are expensive and time-consuming because they require overnight evaluation at sleep laboratories with dedicated systems and attending personnel. To improve such inconveniences of exam and high cost, it is important to develop a simplified method to diagnose the OSA. The motivation of this study is to develop an OSA detection algorithm, which uses only electrocardiogram (ECG) signal. The procedures of this algorithm include three parts. At first, ECG signals are preprocessed by discrete wavelet transform (DWT) method in order to detect the R peck by removing the base line wander and power line noise. Based on the R peak position, the 126 features were generated from the ECG-derived respiration (EDR) signal using time domain and frequency analysis. The range scaling method can be applied afterward to normalize all features in order to minimize the effect of large range variation from the value of each feature. At last, a classification method called support vector machine (SVM) is used to classify if the subject shows OSA symptom or not in each minute. The 10- fold cross-validation method is applied to select the best SVM parameter for classification. By combining all the classification results, one can determine if the subject is normal or apnea. The accuracy of 88.29%, the sensitivity of 92.90% and the specificity value of 86.48%can be seen from the Apnea-ECG database using the minute by minute performance based algorithm. Beside, the accuracy of 100%, the sensitivity of 100% and the specificity value of 100% also can be seen from the database using the subject based performance. The goal of this study to reduce both the detection time and cost is accomplished.

碩士
元智大學
資訊管理學系
97
Abstract Heart attack is one of the major diseases leading to death based on recent reports. Therefore, it is crucial to develop a computer-assisted algorithm to improve conventional diagnoses of heart diseases. 12-lead ECG is a frequently-used diagnostic tool with the advantages of non-invasive measurement and convenient acquisition. Objective : The major objective of this study is to develop a hidden markov model (HMM) that can recognize the ECG features of 12-lead ECG. The HMM model then can be used to identify myocardial infarction that is a life-threaten disease and commonly-seen in clinical practice. Method : The 12-lead ECGs confirmed as myocardial infarction were acquired from clinically-used Philips XML-ECG. The waveforms in collected XML-ECG files were extracted and then processed to get clean ECG waveform data. The waveform data in various leads were used to build a HMM model. The HMM model using mixed Gaussian functions to model waveform pattern in time domain. Result : By using Maximum Likelihood Estimation in the process of HMM training, the optimal HMM model representing anterior myocardial infarction was evaluated. Results indicated that anterior myocardial infarction can be represented by two Gaussian mixture and twelve states in HMM. Conclusion : In sum, a computer-assisted myocardial infarction detector can be facilitated with the use of HMM. Keywords: Hidden Markov Model, 12-lead ECG, Myocardial Infarction

碩士
國立中正大學
電機工程研究所
100
We proposed an emotion recognition system based on electrocardiogram (ECG) in this study which could recognize seven kinds of emotions, including “Neutral”, “Happy”, “Stress”, “Sad”, “Anger”, “Disgust”, “Surprise”. These emotions have been defined as basic emotions in the psychology. Our user-independent emotion recognition system can be divided into five parts: data acquisition (physiological signals), feature calculation, feature extraction, feature selection, and classification. In the physiological signal acquisition part, we recorded ECG from10 participants who watched video films of two to four minutes in length for stimulating distinct emotions in a quiet room. Seven categories of features were calculated from the ECG segment, including time-domain features, HRV features, Poincare plot regional features, baseline features, nonlinear features, frequency-domain features, and waveform features. In our research, the performance of feature extractors and feature selectors were especially discussed. Two feature extractors, including principal component analysis (PCA) and discriminant analysis (DA) were applied to reduce feature dimensions from mapping the original data to new subspace. Three feature selectors, including sequential backward selection (SBS), sequential forward selection (SFS), and genetic algorithms (GA), proceeded to select useful features and reduce feature dimensions. Finally, the leave-one-out cross-validation was performed in combination with the support vector machine (SVM) classifier to recognize the seven emotions mentioned from above. The emotion recognition system using all the 65 features led to a classification rate of 85.71%. When in the single feature category condition, the best system classification rate was 78.57% with Poincare plot regional features. If we removed Poincare plot regional features from the feature set, the classification rate only led to 70.00%, which was considered that the Poincare plot regional features contributed good performance to our system. In our study, we used those 65 features for further research. When applying the two feature extractors for our system, the classification rates were 77.14% when using DA feature extractor and 72.86% when using PCA feature extractor. On the other side, applying the three feature selectors for the emotion recognition system, when applying the GA feature selector, the classification rate could achieve an average accuracy of 98.57%. Although applying feature selectors to our system led to a high classification rate, the time it took also quite long. As a result, when the designer decided which method to apply, the application purpose became a very important role.

碩士
國立臺灣師範大學
應用電子科技學系
98
The electrocardiogram (ECG) analysis is one of the most important approaches to cardiac arrhythmia detection. Many algorithms have been proposed, however, the recognition rate is still unsatisfactory due to unreliable feature extraction in signal characteristic analysis or poor generalization capability of the classifier. In this paper, we propose a system for cardiac arrhythmia detection in ECGs with adaptive feature selection and modified support vector machines (SVMs). Wavelet transform-based coefficients and signal amplitude/interval parameters are first enumerated as candidates, but only a few specific ones are adaptively selected for the classification of each class pair. A new classifier, which integrates k-means clustering, one-against-one SVMs, and a modified majority voting mechanism, is proposed to further improve the recognition rate for extremely similar classes. By testing the system with more than 100,000 samples in MIT-BIH arrhythmia database, the average recognition rate is 97.72%.

Arrhythmogenic cardiomyopathy is a heart muscle disease clinically characterized by life-threatening ventricular arrhythmias and pathologically by an acquired and progressive dystrophy of the ventricular myocardium with fibrofatty replacement. The clinical manifestations of arrhythmogenic cardiomyopathy vary according to the ‘phenotypic’ stage of the underlying disease process. Since there is no ‘gold standard’ to reach the diagnosis of arrhythmogenic cardiomyopathy, multiple categories of diagnostic information have been combined. Different diagnostic categories include right ventricular morphofunctional abnormalities (by echocardiography and/or angiography and/or cardiovascular magnetic resonance imaging), histopathological features on endomyocardial biopsy, electrocardiogram, arrhythmias, and family history, including genetics. The diagnostic criteria were revised in 2010 to improve diagnostic sensitivity, but with the important prerequisite of maintaining diagnostic specificity. Quantitative parameters have been put forward and abnormalities are defined based on the comparison with normal subject data. A definite diagnosis of arrhythmogenic cardiomyopathy is achieved when two major, or one major and two minor, or four minor criteria from different categories are met. The main differential diagnoses are idiopathic right ventricular outflow tract tachycardia, myocarditis, sarcoidosis, dilated cardiomyopathy, right ventricular infarction, congenital heart diseases with right ventricular overload, and athlete’s heart. Among diagnostic tools, contrast-enhanced cardiovascular magnetic resonance is playing a major role in detecting subepicardial-midmural left ventricular free wall involvement, even preceding morphofunctional abnormalities. Moreover, electroanatomical mapping is an invasive tool able to detect early right ventricular free wall involvement in terms of low-voltage areas. Both techniques are increasingly used in the diagnostic work-up although are not yet part of diagnostic criteria.

AbstractClinical data is often collected and processed as time series: a sequence of data indexed by successive time points. Such time series can be from sources that are sampled over short time intervals to represent continuous biophysical wave-(one word waveforms) forms such as the voltage measurements representing the electrocardiogram, to measurements that are sampled daily, weekly, yearly, etc. such as patient weight, blood triglyceride levels, etc. When analyzing clinical data or designing biomedical systems for measurements, interventions, or diagnostic aids, it is important to represent the information contained within such time series in a more compact or meaningful form (e.g., noise filtering), amenable to interpretation by a human or computer. This process is known as feature extraction. This chapter will discuss some fundamental techniques for extracting features from time series representing general forms of clinical data.

AbstractFor medical treatments, pain is often measured by self-report. However, the current subjective pain assessment highly depends on the patient’s response and is therefore unreliable. In this paper, we propose a physiological-signals-based objective pain recognition method that can extract new features, which have never been discovered in pain detection, from electrodermal activity (EDA) and electrocardiogram (ECG) signals. To discriminate the absence and presence of pain, we establish four classification tasks and build four tree-based classifiers, including Random Forest, Adaptive Boosting (AdaBoost), eXtreme Gradient Boosting (XGBoost), and TabNet. The comparative experiments demonstrate that our method using the EDA and ECG features yields accurate classification results. Furthermore, the TabNet achieves a large accuracy improvement using our ECG features and a classification accuracy of 94.51% using the features selected from the fusion of the two signals.