Sleep disordered breathing (SDB) represents an important risk factor for the health. The clinical context is of increasing severity, ranging from simple snoring, passing through the upper airway resistance syndrome (UARS) to more serious obstructive sleep apnea syndrome (OSAS). The most common SDB symptoms include snoring at night, headache upon awakening, daytime sleepiness and decrease cognitive performance. In addition, these symptoms may produce even more serious consequences, including social problems in the work place and traffic accidents. Of great importance is the case of OSAS. The failure in recognize this condition is particularly serious considered that the major cases of OSA produce arterial hypertension and cardiovascular diseases. For all this reasons, the sleep apnea syndrome is a serious problem of social and medical interest and diagnostic strategies are therefore necessary to reduce their prevalence and limit their consequences. The gold standard procedure to analyze sleep and diagnosis sleep disorders is a sleep study performed in a sleep laboratory. Dedicated personnel, adequate infrastructure and special acquisition systems make the diagnostic procedure cumbersome and expensive. Polysomnography, defined as the gold standard in the analysis of sleep, indeed, requires the recording of different signals. Typically, a full night’s sleep is observed before a diagnosis is reached and in some subjects even a second night’s recording is required. Due to the high cost and to the small number of laboratories equipped for the sleep study, it is estimated that sleep apnea is widely under diagnosed. From this picture of objective difficulties emerges the motivation of new diagnostic techniques, more accessible and easy to implement without the need for specialized laboratory, at least in a screening phase of diagnosis. Although sleep apnea is a breathing event, its effects can clearly be seen by other peripheral systems such as the cardiovascular system. In this direction there is a close relationship between the electrocardiographic signal (ECG) and the respiratory signal. Therefore evaluate the respiratory signal from the ECG, and in particular through a portable minimally invasive device as a Holter recorder, can be a viable alternative. The use of the Holter is needed to have a continuous recording of the ECG during the sleep hours. The aim of this thesis is to analyze various ECG derived respiratory signals (EDR) and sleep apnea through the Holter ECG signal. On this surrogate signals of the breathing has been implemented an apnea recognition algorithm. This work is part of a more extensive project that involves the Politecnico di Milano with the Departments of Cardiology 3 and Cardiology 4 and the Center for Epilepsy Surgery, Center for Sleep Medicine at Niguarda Ca’ Granda Hospital in Milan. The aim of this project is to identify with a simple test, non-invasive, inexpensive and widely used, those subjects at risk of OSAS that are then directed to perform polysomnography, more complex and expensive, so to confirm any diagnosis of sleep apnea syndrome. The signals analyzed in this thesis have been obtained by recording Holter ECG signals and polysomnography in a population of patients with obstructive sleep apnea. The thesis has four main chapters. The first chapter presents the clinical-physiological interaction between the respiratory and cardiovascular systems. In this part are examined studies on the identification of the causes and effects of the interaction between breathing and ECG. The second chapter is a literature review on several methods for estimating the breath from the ECG. This constitutes the analysis of the state of art algorithms currently implemented to estimate the breath. Three categories of algorithm has been studied. A firs category considered is that based on the study of the variability in the morphology and alignment of the vectorcardiographic loop (VCG) and in the direction and measure of the mean cardiac electrical axis. A second category of algorithms is related to the modulations in the amplitude of the R peaks, R-S amplitude and baseline wander of the ECG trace. A further method of this second category, which has proved very effective relative to its simplicity of implementation, is that of the breath estimated by the area below to each QRS complex of the ECG. The third category of algorithms presented is relative to the heart rate variability (HRV) and the feature that explain it is the inter-beat distance (RR intervals). The third chapter describes the methods adopted in this work to estimate the breath. In particular two types of EDR has been implemented. The first one from the three ECG orthogonal leads and the second one from a single ECG lead. For the case involving the three ECG orthogonal leads has been used a method based on the evaluation of the VCG loop in correspondence of each QRS complex followed by an evaluation using the principal component analysis (PCA) of a set of parameters containing three spatial coordinates of the centers of gravity and three directions of the three inertial axis. For the case related to the single ECG lead have been implemented in the order (1) the method based on the areas below the QRS complex of each beat, (2) the amplitude of the R peaks, (3) the R-S amplitude, (4) the baseline wander of the ECG and (5) the RR intervals. Finally, the four chapter describes the algorithm of apnea detection processed on the EDR signals and the results are presented. The apnea detection algorithm compares the amplitude of breathing with the last reference amplitude and by means of a comparison with a threshold below which the signal should be for at least a time period of ten seconds, it performs the apnea detection. The results obtained in this study were compared with quantitate signals of breathing (thorax plethysmographic signal) and lead to an encouraging use of this technique for effective screening of subjects with symptoms and signs peculiar of the pathology.
I disturbi respiratori durante il sonno, o sleep disordered breathing (SDB), costituiscono un fattore di rischio importante per la salute. Il quadro clinico risulta di gravitá crescente partendo dal semplice russamento (simple snoring), passando dalla sindrome da aumentate resistenze delle alte vie aeree (upper airway resistance syndrome, UARS) fino a giungere alla sindrome delle apnee ostruttive durante il sonno (obstructive sleep apnea syndrome, OSAS). I sintomi piú comuni di tali disturbi comprendono il russamento notturno, cefalea al risveglio, la sonnolenza diurna e la diminuzione delle performance cognitive. Inoltre, questi sintomi, possono generare conseguenze anche serie che trovano drammatiche implicazioni in problemi sociali nel posto di lavoro e incidenti automobilistici. Di importante rilevanza é il caso dell’OSAS. Il mancato riconoscimento di questa condizione é particolarmente grave dato che casi importanti di OSA generano ipertensione arteriosa, malattie cardiovascolari. Per tutte queste ragioni, la sindrome delle apnee notturne rappresenta un serio problema di carattere socio-sanitario e pertanto sono necessarie strategie diagnostiche volte a ridurne la prevalenza e a limitarne le conseguenze. La procedura di riferimento per l’analisi del sonno e la diagnosi degli eventi apneici é uno studio condotto in laboratori del sonno. Personale specializzato, infrasturtture adeguate e speciali sistemi di acquisizione rendono la procedura di diagnosi poco agevole e costosa. La polisonnografia, esame definito di riferimento nell’analisi del sonno, richiede, infatti, la registrazione di diversi segnali. Tipicamente, prima di raggiungere una diagnosi si osservano i dati relativi ad un’intera notte e in alcuni soggetti spesso risulta necessaria addirittura una seconda notte di registrazione. A causa degli ingenti costi e della relativa scarsitá di laboratori attrezzati per lo studio del sonno, é stato stimato che le apnee durante il sonno sono ampiamente sotto-diagnosticate. Da questo quadro di difficoltá oggettive emerge la motivazione di nuove tecniche diagnostiche piú semplici e accessibili che facciano a meno della necessitá di laboratori specializzati, almeno in una fase inziale di diagnosi. Sebbene l’apnea sia un evento respiratorio, i suoi effetti possono chiaramente essere osservati da altri sistemi periferici come il sistema cardiovascolare. In questo senso c’é una stretta relazione tra il segnale elettrocardiografico (ECG) e quello respiratorio. Dunque ottenere il segnale respiratorio da ECG, e in particolare mediante un dispositivo portatile e minimamente invasivo come un registratore Holter, puó costituire una valida alternativa. Inoltre l’utilizzo dell’Holter risulta necessario per permettere una registrazione continua dell’ECG durante le ore del sonno. L’obiettivo di questo lavoro di tesi é di analizzare diverse stime del segnale respiratorio e delle apnee notturne attraverso la rielaborazione del segnale ECG Holter mediante l’implementazione di due tipologie di algoritmi: la prima su segnali ECG a tre derivazioni ortogonali e la seconda su singola derivazione del segnale ECG. Su quete due tipologie di segnali surrogati del respiro é stato implementato un algoritmo di riconoscimento delle apnee. Questo lavoro si inserisce all’interno di un progetto piú ampio che vede coinvolto il Politecnico di Milano con i Dipartimenti di Cardiologia 3 e Cardiologia 4 e il Centro per la Chirurgia dell’Epilessia, Centro di Medicina del Sonno dell’Ospedale Niguarda Ca’ Granda di Milano. Lo scopo di tale progetto é di individuare con un esame semplice, non invasivo, poco costoso e molto diffuso, i soggetti a rischio di OSAS che vengano quindi indirizzati ad effettuare la polisonnografia, piú complessa e costosa, per confermare quindi l’eventuale diagnosi di sindrome delle apnee del sonno. I segnali analizzati in questa tesi sono stati ottenuti tramite registrazione elettrocardiografica secondo Holter e segnale polisonnografico in una popolazione di pazienti affetti da apnee ostruttive del sonno. La tesi si sviluppa in quattro capitoli principali. Il primo capitolo presenta l’aspetto clinico-fisiologico dell’interazione fra il sistema respiratorio e quello cardiovascolare. In questa parte vengono esaminati studi sull’identificazione delle cause ed effetti dell’interazione fra respiro ed ECG. Il secondo capitolo é una revisione della letteratura su metodi per la stima del respiro da segnale ECG. Questa parte constituisce l’analisi dello stato dell’arte di algoritmi ad oggi implementati per la stima del respiro. Sono state studiate tre categorie di algoritmi. Una prima categoria di algoritmi presi in considerazione é quella basata sullo studio della variabilitá nella morfologia e nell’allineamento dei loop vettorcardiografici e nella direzione e misura dell’asse elettrico cardiaco medio. Una seconda categoria di algoritmi presenti in letteratura é realtiva alle modulazioni in ampiezza dei picchi R, ampiezza R-S e della linea di base del tracciato ECG. Un ulteriore metodo di questa seconda categoria, rivelatosi molto efficace in relazione alla sua relativa semplicitá di implementazione, é quello del respiro stimato mediante le aree sottese ai complessi QRS dell’ECG. La terza categoria di algoritmi presentati é invece relativa alla variabilitá cardiaca (HRV – Heart Rate Variability) che vede come feature la differenza interbattito RR. Il terzo capitolo entra nel merito della descrizione dettagliata dei metodi implementati per la stima del respiro. In particolare vengono implementate due tipologie di derivazione di segnale del respiro. La prima da segnale ECG a tre derivazioni ortogonali e la seconda da segnale ECG di tipo a singola derivazione. Per il caso realtivo alle tre derivazioni ortoganali é stato utilizzato l’algoritmo basato sulla valutazione del loop vectorcardiografico in corrispondeza di ogni complesso QRS seguito dalla valutazione tramite analisi della componente principale di un set di parametri contenenti le tre coordinate spaziali dei centri di gravitá e le tre direzioni nello spazio dei 3 assi di inerzia. Per il caso relativo alla singola derivazioni sono stati implementati nell’ordine (1) l’algoritmo basato sulle aree sottese al complesso QRS di ciascun battito, (2) l’ampiezza dei picchi R, (3) l’ampiezza R-S, (4) l’andamento della linea di base dell’ECG e (5) gli intervalli RR. In conclusione, nel quarto capitolo viene descritto l’algoritmo di detezione delle apnee processato sui segnali derivati del respiro e vengono presentati i risultati ottenuti. L’algoritmo di detezione delle apnee confronta le ampiezze degli atti respiratori con l’ultima ampiezza di riferimento e tramite un confronto con una soglia sotto la quale il segnale deve trovarsi per almeno un tempo di dieci secondi, si effettua la detezione delle apnee. I risultati ottenuti in questo lavoro sono stati confrontati con i segnali quantitativi del respiro (segnale pletismografico del torace) e fanno emergere un incoraggiante utilizzo di questa tecnica per uno screening efficace di soggetti che presentano sintomi riconosciuti tra quelli della patologia.
Screening for sleep related breathing disorders form ECG Holter respiratory signals
BRUNO, EMANUELE
2014/2015
Abstract
Sleep disordered breathing (SDB) represents an important risk factor for the health. The clinical context is of increasing severity, ranging from simple snoring, passing through the upper airway resistance syndrome (UARS) to more serious obstructive sleep apnea syndrome (OSAS). The most common SDB symptoms include snoring at night, headache upon awakening, daytime sleepiness and decrease cognitive performance. In addition, these symptoms may produce even more serious consequences, including social problems in the work place and traffic accidents. Of great importance is the case of OSAS. The failure in recognize this condition is particularly serious considered that the major cases of OSA produce arterial hypertension and cardiovascular diseases. For all this reasons, the sleep apnea syndrome is a serious problem of social and medical interest and diagnostic strategies are therefore necessary to reduce their prevalence and limit their consequences. The gold standard procedure to analyze sleep and diagnosis sleep disorders is a sleep study performed in a sleep laboratory. Dedicated personnel, adequate infrastructure and special acquisition systems make the diagnostic procedure cumbersome and expensive. Polysomnography, defined as the gold standard in the analysis of sleep, indeed, requires the recording of different signals. Typically, a full night’s sleep is observed before a diagnosis is reached and in some subjects even a second night’s recording is required. Due to the high cost and to the small number of laboratories equipped for the sleep study, it is estimated that sleep apnea is widely under diagnosed. From this picture of objective difficulties emerges the motivation of new diagnostic techniques, more accessible and easy to implement without the need for specialized laboratory, at least in a screening phase of diagnosis. Although sleep apnea is a breathing event, its effects can clearly be seen by other peripheral systems such as the cardiovascular system. In this direction there is a close relationship between the electrocardiographic signal (ECG) and the respiratory signal. Therefore evaluate the respiratory signal from the ECG, and in particular through a portable minimally invasive device as a Holter recorder, can be a viable alternative. The use of the Holter is needed to have a continuous recording of the ECG during the sleep hours. The aim of this thesis is to analyze various ECG derived respiratory signals (EDR) and sleep apnea through the Holter ECG signal. On this surrogate signals of the breathing has been implemented an apnea recognition algorithm. This work is part of a more extensive project that involves the Politecnico di Milano with the Departments of Cardiology 3 and Cardiology 4 and the Center for Epilepsy Surgery, Center for Sleep Medicine at Niguarda Ca’ Granda Hospital in Milan. The aim of this project is to identify with a simple test, non-invasive, inexpensive and widely used, those subjects at risk of OSAS that are then directed to perform polysomnography, more complex and expensive, so to confirm any diagnosis of sleep apnea syndrome. The signals analyzed in this thesis have been obtained by recording Holter ECG signals and polysomnography in a population of patients with obstructive sleep apnea. The thesis has four main chapters. The first chapter presents the clinical-physiological interaction between the respiratory and cardiovascular systems. In this part are examined studies on the identification of the causes and effects of the interaction between breathing and ECG. The second chapter is a literature review on several methods for estimating the breath from the ECG. This constitutes the analysis of the state of art algorithms currently implemented to estimate the breath. Three categories of algorithm has been studied. A firs category considered is that based on the study of the variability in the morphology and alignment of the vectorcardiographic loop (VCG) and in the direction and measure of the mean cardiac electrical axis. A second category of algorithms is related to the modulations in the amplitude of the R peaks, R-S amplitude and baseline wander of the ECG trace. A further method of this second category, which has proved very effective relative to its simplicity of implementation, is that of the breath estimated by the area below to each QRS complex of the ECG. The third category of algorithms presented is relative to the heart rate variability (HRV) and the feature that explain it is the inter-beat distance (RR intervals). The third chapter describes the methods adopted in this work to estimate the breath. In particular two types of EDR has been implemented. The first one from the three ECG orthogonal leads and the second one from a single ECG lead. For the case involving the three ECG orthogonal leads has been used a method based on the evaluation of the VCG loop in correspondence of each QRS complex followed by an evaluation using the principal component analysis (PCA) of a set of parameters containing three spatial coordinates of the centers of gravity and three directions of the three inertial axis. For the case related to the single ECG lead have been implemented in the order (1) the method based on the areas below the QRS complex of each beat, (2) the amplitude of the R peaks, (3) the R-S amplitude, (4) the baseline wander of the ECG and (5) the RR intervals. Finally, the four chapter describes the algorithm of apnea detection processed on the EDR signals and the results are presented. The apnea detection algorithm compares the amplitude of breathing with the last reference amplitude and by means of a comparison with a threshold below which the signal should be for at least a time period of ten seconds, it performs the apnea detection. The results obtained in this study were compared with quantitate signals of breathing (thorax plethysmographic signal) and lead to an encouraging use of this technique for effective screening of subjects with symptoms and signs peculiar of the pathology.File | Dimensione | Formato | |
---|---|---|---|
2015_12_Bruno.pdf
accessibile in internet solo dagli utenti autorizzati
Descrizione: Testo della tesi
Dimensione
13.82 MB
Formato
Adobe PDF
|
13.82 MB | Adobe PDF | Visualizza/Apri |
I documenti in POLITesi sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
https://hdl.handle.net/10589/114381