شناسایی آریتمی به کمک الکتروکاردیوگرام 2 لید با استفاده از شبکه عصبی مصنوعی و ماشین بردار پشتیبان برای یک دستگاه مانیتور ECG دستی

Arrhythmia Identification with Two-Lead Electrocardiograms Using Artificial Neural Networks and Support Vector Machines for a Portable ECG Monitor System

Abstract

An automatic configuration that can detect the position of R-waves, classify the normal sinus rhythm (NSR) and other four arrhythmic types from the continuous ECG signals obtained from the MIT-BIH arrhythmia database is proposed. In this configuration, a support vector machine (SVM) was used to detect and mark the ECG heartbeats with raw signals and differential signals of a lead ECG. An algorithm based on the extracted markers segments waveforms of Lead II and V1 of the ECG as the pattern classification features. A self-constructing neural fuzzy inference network (SoNFIN) was used to classify NSR and four arrhythmia types, including premature ventricular contraction (PVC), premature atrium contraction (PAC), left bundle branch block (LBBB), and right bundle branch block (RBBB). In a real scenario, the classification results show the accuracy achieved is 96.4%. This performance is suitable for a portable ECG monitor system for home care purposes.

1 Introduction

Telemedicine has been widely studied recently. In past research, allowing congestive heart failure patients to monitor their condition at home offered great economic advantages. Electrocardiograms (ECGs) are an important tool that provide useful information about the functional status of the heart. An automated method that accurately diagnoses cardiac diseases through the analysis of ECG signals is critical in healthcare [1], especially for real-time processing. Past research has addressed the problems of heart rate detection and classification of cardiac rhythms. The heart rate signal detects the QRS wave of the ECG and calculates inter-beat intervals [2–9]. The classification of cardiac rhythms is based on the detection of the different types of arrhythmia from the ECG waveforms [10–13].

However, ECG signals have coupling noises, due to factors such as 50/60 Hz power line signals, the baseline drift caused by patient breathing, bad electrodes, improper electrode location, or electromyograms. These noises result in false QRS wave detections. Thus, some studies have compared the robust performance of different algorithms for QRS wave detection [2]. Widrow et al. applied the adaptive filter to reduce noises that resulted from 60 Hz power lines and baseline drift, and then detect the QRS wave [14]. Pan and Tompkins designed a digital filter to reduce the noise and used a dynamic threshold to detect the QRS wave [4]. Trahanias used the mathematical morphology of the QRS complex to detect heart rates [5]. Chang used the ensemble empirical model decomposition to reduce noises in arrhythmia ECGs [15]. Fan used approximate entropy (ApEn) and Lempel-Ziv complexity as a nonlinear quantification to measure the depth of anaesthesia [16]. In these studies, the normal sinus ECG signal added different noise types and energy was used to evaluate the performance of these algorithms. Several researchers have extracted the features of ECG waveforms to detect the QRS complexes based on the arrhythmia database. Li et al. proposed the wavelet transforms method for detecting the QRS complex from high P or T waves, noise, and baseline drift [6]. Yeh and Wang proposed the difference operation method to detect the QRS complex waves [8]. Mehta and Lingayat used the support vector machine (SVM) method to detect the QRS complexes from a 12-leads ECG [9]. They also used the K-mean algorithm for the detection of QRS complexes in ECG signals [17].

Arrhythmia can be defined as either an irregular single heartbeat or a group of heartbeats. Some classification techniques are based on the ECG beat-by-beat classification with each beat being classified into several different arrhythmic beat types. These include artificial neural networks [11], fuzzy neural networks [18], Hermite functions combined with self-organizing maps [19], and wavelet analysis combined with radial basis function neural networks [20]. In these methods, the ECG waveform of each beat was picked up manually and different features were extracted to classify the arrhythmic types. Tsipouras et al. used the RR-interval signal to classify certain types of arrhythmia based on a group of heartbeats [12]. All the above methods have high classification accuracies that were obtained based on the complete morphology of the ECG or the correct RR-interval that was detected manually.

In this study, we propose an automatic configuration integrating digital signal processing and an artificial intelligence method to detect the position of heartbeats and recognize these heartbeats as belonging to the normal sinus rhythm (NSR) or four arrhythmic types. The four arrhythmic types are premature ventricular contraction (PVC), premature atrium contraction (PAC), left bundle branch block (LBBB), and right bundle branch block (RBBB). ECG signals are provided by the MIT-BIH Arrhythmia Database [21]. This automatic configuration had three steps, as follows:

1. The Lead II signals were normalized and filtered to reduce the coupled noise (Section 2.2).

2. The positions of QRS-complexes in Lead II were detected and marked via a well-trained SVM. Two waveforms of each heartbeat in Lead II and V1 were individually extracted according the markers in Lead II (Section 2.3).

3. The extracted waveform was used as a feature to recognize the arrhythmic type of a heartbeat. In this configuration, a self-constructing neural fuzzy inference network (SoNFIN) was used to recognize the arrhythmic type of the heartbeat using the raw Lead II and V1 signals (Section 2.4).

Moreover, the heartbeat detection accuracy has been increased by the SoNFIN classification results.

مقالات انگلیسی و ترجمه های فارسی رایگان با فرمت PDF آماده دانلود رایگان میباشند

همچنین ترجمه مقاله با فرمت ورد نیز قابل خریداری و دانلود میباشد

مقاله انگلیسی رایگان (PDF)

ترجمه فارسی رایگان (PDF)

خرید ترجمه با فرمت ورد

فهرست مقاله:

چکیده
1- مقدمه
2- مرور منابع
3- الگوريتم پيشنهادی
الف) پيش نيازها
ب) چولگي
ج) الگوريتم
4- شبيه سازي و آزمايش داده های واقعي
5- نتيجه گيری

بخشی از ترجمه فارسی مقاله:

1- مقدمه

مطالعات نشان مي دهد كه مهم ترين منبع استرس مادر در دوران بارداري سالم بودن جنين است. هر سال حدود هشت نوزاد از بين هر هزار نوزاد با برخي نارسايي هاي قبلي مادرزادي متولد مي شوند. نارسايي مي تواند چنان ناچيز باشد كه در عمل و بعد از ساليان زيادي بعد از تولد،  نوزاد سالم به نظر برسد يا چنان مهم و حاد باشد كه زندگي نوزاد در خطر باشد.

يك راه آگاهي يافتن از سلامت جنين در دوران بارداري اين است كه سيگنال هاي الكتروكاردياگرامي ثبت شده توسط روش  غيرتهاجمي در نظر گرفته شود. در اين روش، الكتروكاردياگرام ها از شكم مادر ثبت مي شوند. سيگنال ثبت شده تركيبي از الكتروكاردياگرام هاي مادر عبوري از سينه تا شكم، الكتروكاردياگرام جنين و نويز مي باشد.بيشتر نارسايي هاي قلبي داراي برخي علایم خود را در سيگنال هاي الكتريكي قلبي نشان می دهند كه توسط الكتروكاردياگرامي ثبت شده اند و در ارتباط با آنها اين اعتقاد وجود دارد كه اطلاعات زيادي در مقايسه با روش-هاي سونوگرافي متداول داشته باشند. با اين وجود، به دليل SNR كم الكتروكاردياگرام جنين كه از سطح بدن مادران ثبت مي شود، ما به الگوريتمي نياز داريم تا الكتروكاردياگرام را از بين اين سيگنال هاي ثبت شده اسخراج نمايد.اخيرا، محققين دريافتند كه اين مسئله مي تواند به صورت جداسازي منبع كور (BSS) مدلسازي شود. جداسازي منبع كور (يا تحليل مولفه مستقل، ICA) تمام سيگنال هاي منبع را از تعداد زيادي سيگنال سنسوري مشاهده شده كه مي توانست زمان زيادي طول بكشد، استخراج مي نمايد  و تنها تعداد خيلي كمي از سيگنال هاي منبع مورد نظر هستند. براي اين كاربرد، تكنيك ديگر به نام استخراج سيگنال كور (BSE) كانديد قدرتمندي است، زيرا الگوريتم يادگيري BSE مي تواند يك سيگنال منبع تكي را از يك تركيب خطي سيگنال هاي منبع استخراج نمايد. بنابراين ما در جست وجوي دست يابي به سه هدف زير هستيم:

استخراج تنها مولفه سيگنال مورد نظر (FECG) به عنوان يك خروجي،

بهبود كيفيت سيگنال استخراج شده با افزايش SNRsvdو SNRcor،

كاهش زمان محاسباتي به منظور ايجاد يك الگوريتم آني و بلادرنگ.

به منظور دست يابي به اهداف ذكر شده فوق، ما يك الگوريتم پيشنهاد مي كنيم كه از گستره مقادير چولگي سيگنال مورد نظر (FECG) استفاده مي نمايد. اعتبار و درستي روش ما بر روي داده هاي ECG واقعي تست شده است. مابقي مطالب مقاله به صورت زير سامان دهي شده است:بعد از يك بازبيني مختصر مطالعات پيشين در بخش 2، بخش 3 جزئيات روش پيشنهادي ما را فراهم مي كند. بخش 4 نتايج تجربي را گزارش مي كند و در نهايت بخش 5 دربرگيرنده نتايج پاياني خواهد بود.

2- مرور منابع

مطالعات گسترده اي در ارتباط با استخراج نوار قلب جنين هدايت شده اند. مبحث اينجا قصد ارائه بازبيني جامع و كامل از اين موضوع را ندارد، اما برخي از مهم ترين مطالعات اين زمينه را معرفي مي نمايد. در [1]، تكنيك فيلتر تطبيق پذير براي حذف نوار قلب مادر و دست يابي به نوار قلب جنين استفاده گرديد. مولفين دو مجموعه از الكترودها را مورد استفاده قرار داده اند، يك مجموعه بر روي شكم مادر (ورودي هاي اوليه) جاي داده شد و مجموعه ديگر روي سينه مادر قرار گرفت (ورودي هاي مرجع). با داشتن اين سيگنال ها به عنوان ورودي به فيلتر تطبيق پذير، سيگنال خطا مي تواند براي نمايش نوار قلب جنين استخراج شده ايجاد شوند. در [2]، روشي با الگوريتم ژنتيك كه داراي ساختار مشابهي با روش فوق بود، پيشنهاد گرديد. اين روش ها به اندازه كافي قدرتمند نبودند تا به منصه ظهور برسند. زيرا اگر دامنه نويز پس زمينه بزرگ تر از ضربان قلب جنين بود، سيگنال خطاي منتجه دربرگيرنده نوار قلب جنين نخواهد بود و همچنين در حالتي كه نوار قلب مادر و جنين دچار هم پوشاني بودند، هر دو روش براي استخراج نوار قلب جنين ناموفق عمل مي كردند.در [3]، مفهوم بردارهاي انرژي منفرد متعامد براي استخراج FECG از AECG مورد استفاده قرار گرفت. اين روش به طور گسترده در محيط هاي كلينيكي استفاده نشد، زيرا سيگنال مورد نظر تنها زماني مي توانست به طور موفقيت آميز استخراج شود كه الكترودها در موقعيتي نصب شود كه بردارهاي انرژي منفرد MECG و FECG نسبت به هم متعامد باشند.در [4]، باروس و كيچوكي يك الگوريتم جداسازي منبع نيمه-كور براي حل مسئله استخراج نوار قلب جنين پيشنهاد كردند. اين الگوريتم به اطلاعات پيشين در رابطه با تابع هم بستگي منابع اوليه نياز دارد تا سيگنال مورد نظر (FECG) استخراج شود. آنها فرض نكردند كه منابع به طور آماري مستقل باشد اما فرض كردند كه منابع ساختار زماني دارند و تابع هم بستگي مختلفي دارند. مسئله اصلي در اين روش اين است كه اگر نرخ ضربان قلب جنين متغير باشد، كه بيان كننده اين موضوع است كه جنين سالم نيست، تخمين پيشين تابع هم بستگي نوار قلب جنين ممكن است مناسب نباشد.همچنين در [5] ژانگ و ژانگ يي يك الگوريتم استخراج منبع كور (BSE) براي استخراج يك سيگنال منبعي از سيگنال هاي مشاهده شده پيشنهاد كردند. الگوريتم آنها به يك آگاهي آماري قبلي از سيگنال مورد نظر نياز داشت. اين بدين معني است كه با دانستن رنجي كه مقدار اوج نمودار سيگنال در آن دارا بود، روش قادر بود تا سيگنال مورد نظر را استخراج نمايد.در [13]، مينگ و يو-لين يك الگوريتم را پيشنهاد كردند كه تركيبي از تناوب پذيري و مقدار اوج نموداري بود كه به ترتيب توسط باروس [4]، و ژانگ [5] پيشنهاد شده بودند. همان طور كه در بالا ذكر شد، در اين حالت زماني كه جنين سالم نيست، تخمين پيشين تابع هم بستگي نوار قلب جنين ممكن است مناسب نباشد.در اين مقاله، ما مقدار کشیدگی نمودار را با مقدار چولگي جايگزين كرده ايم كه ما را متقاعد مي سازد تا بهبودهاي ديگري نسبت به روش آنها اعمال نمائيم. اين بهبودها در بخش هاي بعدي توضيح داده خواهد شد. با اين جايگزيني، ما كيفيت سيگنال مورد نظر را بهبود بخشيديم و مقدار SNRsvd و SNRcor افزايش يافت. همچنين بارمحاسباتي الگوريتم پيشنهادي كاهش يافته است.

بخشی از مقاله انگلیسی:

I. INTRODUCTION

Studies show that the most important source of mother’s stress in pregnancy is because of fetus’s health condition. Every year about eight out of one thousand babies are born with some form of congenital heart defects. The defect may be so slight that the baby appears healthy for many years after birth, or so important that his/her life is in immediate danger. One way of knowing about the fetus’s health condition in pregnancy is to consider the electrocardiographic signal recorded using non-invasive method. In this method, electrocardiograms are recorded from the mother’s abdomen. Recorded signal is a combination of mother’s electrocardiogram after travelling from the chest to abdomen, fetus’s electrocardiogram and noise. Most cardiac defects have some manifestation in morphology of cardiac electrical signals, which are recorded by electrocardiography and are believed to contain much more information as compared with conventional sonographic methods. However, due to the low SNR of fetal electrocardiogram (ECG) recorded from the maternal body surface, we need an algorithm to extract fetal electrocardiogram among these recorded signals. Recently, researchers found that the problem can be modelled as the blind source separation (BSS). Blind Source Separation (or, Independent Component Analysis, ICA), extracts all the source signals from a large number of observed sensor signals could take a long time and only a very few source signals are subjects of interest. For this application, another technique, blind (semi-blind) signal extraction (BSE) is a powerful candidate, since the BSE learning algorithms can extract a single source signal from a linear mixture of source signals. Therefore we are seeking three following targets: Extracting only desired signal (FECG) as an output Improving quality of extracted signal by increasing SNRsvd and SNRcor. Decreasing the computational time in order to making a real-time algorithm. For achieving targets mentioned above we proposed an algorithm that using the range of skewness value of desired signal (FECG). Validity of our approach has been tested on real-world ECG data. The remainder of paper is organised as follow: After a brief review of literatures in section 2, section 3 details the approach we proposed. Section 4 reports the experimental results and finally section 5 contains concluding remarks.

II. LITERATURE REVIEW

Extensive studies have been conducted related to fetal ECG extraction. The discussion here is not intended to be comprehensive, but highlights some the most important one of them. In [1], adaptive filter technique was used to cancel the mother ECG and obtain the fetal ECG. They used two set of electrodes, one set place on abdomen of mother (primary inputs), and the other placed on the chest of the mother (reference inputs). Having these signals as an input to adaptive filter, the error signal can be made to represent the extracted fetal ECG. In [2], the method proposed by genetic algorithm that has a similar structure as method mentioned above. These methods are not robust enough to be used, because if the amplitude of background noise is greater than the fetal heart beat, the resulting error signal will not contain fetal ECG and when both the mother and fetal ECG overlapped in each other both methods will fail to extract the fetal ECG. In [3], concept of singular energy vectors orthogonality was used to extract FECG form AECG. This method was not exploited widely in clinical environment, because the desired signal can be extracted successfully only if electrodes placed in a position that singular energy vectors of MECG and FECG are orthogonal to each other. In [4], Barros and Cichocki proposed a semi-blind source separation algorithm to solve the fetal ECG extraction problem. This algorithm requires a priori information about the autocorrelation function of primary sources, to extracts the desired signal (FECG). They do not assume the sources to be statistically independent but they assume that sources have temporal structure and have different autocorrelation function. The main problem with this method is that if there is fetal heart rate variability, as in the case when the fetus is not healthy, the priori estimate of the autocorrelation function of the fetal ECG may not be appropriate. Also in [5], Zhang and Zhang Yi proposed a blind source extraction (BSE) algorithm to extract a source signal from observed signals. Their algorithm needs a preknowledge about desired signal statistics. It means that by knowing the range in which the kurtosis value of the signals lies, the method can extract the desired signal. In [13], Ming and Yu-lin proposed an algorithm that is combining of periodicity and kurtosis that proposed by barros [4], and Zhang [5], respectively. As mentioned above, in the case when the fetus is not healthy , the priori estimate of the autocorrelation function of the fetal ECG may not be appropriate. In this paper we substituted kurtosis value with skewness value that persuaded us to applying some other modification to their method. These modifications will be explained in following section. By this substitution, we improved the quality of desired signal and increased SNRsvd and SNRcor. Also computational cost of proposed algorithm has been decreased.