Calvo, M.; Gomis, P.; Romero, D.; Le Rolle, V.; Béhar, N.; Mabo, P.; Hernández, A. Physiological measurement Vol. 38, num. 2, p. 387-396 DOI: 10.1088/1361-6579/aa513c Data de publicació: 2017-02-01 Article en revista
Symptoms such as ventricular arrhythmias in Brugada syndrome (BS) typically occur at rest, especially during sleep, suggesting that the autonomic nervous system (ANS) function may be relevant in the arrhythmogenesis of the disease. The aim of this work was to assess the ANS response captured by a non-linear heart rate variability (HRV) measure in 69 patients diagnosed with BS, who underwent a standardized physical stress test. Heart rate complexity (HRC) was evaluated by the power-law scaling analysis (Beta slope) during rest, exercise, recovery and rest post-recovery, in order to discriminate between symptomatic and asymptomatic BS patients. Symptomatic patients showed a significant reduction in HRC in comparison to asymptomatic subjects, after exertion (p=0.015); during the whole recovery period (p=0.023), and, in particular within the passive recovery phase (p=0.025), as well as during rest post-recovery (p=0.022). Based on these results, symptoms could be associated with a lower ANS complexity during the stress test stages where parasympathetic activity is predominant. Therefore, the proposed HRV indicators could be of help in the risk stratification of asymptomatic patients.
La isquemia de miocardio, consecuencia de la enfermedad de las arterias coronarias, es el principal problema de salud que conduce a infarto de miocardio, arritmias y muerte. El sistema nervioso autónomo (SNA) regula la mayoría de funciones del organismo, entre éstas la frecuencia cardiaca. La caracterización de la complejidad de la respuesta del SNA con técnicas multifractales sobre la serie de intervalos RR, y su análisis antes y durante isquemia de miocardio inducida es un aporte valioso para la comprensión de los mecanismos autónomos de control cardiaco. Sin embargo, hasta el momento, el uso de tales técnicas en el diagnóstico clínico de enfermedades como la isquemia de miocardio y otras ha sido prácticamente nulo. Así mismo, el uso de índices derivados de los intervalos de despolarización y repolarización ventricular en la señal de electrocardiograma (ECG) es una de las vías habituales de detección de enfermedades isquémicas cardiacas.El trabajo desarrollado en esta tesis tiene tres objetivos fundamentales. El primero es caracterizar la respuesta no lineal del SNA durante isquemia de miocardio inducida a través del análisis de la dinámica no lineal presente en las fluctuaciones de la serie de intevalos RR de corta duración, utilizando métodos de análisis fractal (mono y multifractal) y técnicas de surrogate data. El segundo objetivo es analizar índices de repolarización y despolarización ventricular a partir de la señal de ECG de alta resolución para identificar cuáles de éstos describen mejor los episodios de isquemia de miocardio en términos de la potencia estadística discriminatoria de cada uno por separado, del tamaño del efecto, así como su grado de interacción. Por último, costruir un modelo altamente robusto de predicción de isquemia e infarto de miocardio (MI) con métodos de aprendizaje automático basados en las medidas anteriormente identificadas.La base de datos empleada en esta tesis contiene señales de ECG de pacientes admitidos a angioplastia coronaria electiva para aumentar la luz de las arterias coronarias obstruidas. Los síntomas observados durante la oclusión coronaria completa producida por el inflado del balón de angioplastia son similares a los encontrados en pacientes que sufren MI, de ahí que la técnica es un modelo excelente para estudiar la isquemia y el MI. El procedimiento practicado aquí es único pues la duración de la oclusión ha sido más larga que lo habitual, lo cual permite estudiar episodios de isquemia transitoria de miocardio y la fase temprana de un MI.En general, los resultados obtenidos muestran un aumento en la complejidad multifractal y la no linealidad del SNA durante isquemia, lo cual ha sido interpretado como un mecanismo de adaptación beneficioso para incrementar el flujo de sangre hacia las zonas afectadas del miocardio. El análisis desde la perspectiva de la evaluación de los cambios autónomos a través de la serie de intervalos RR de corta duración representa un enfoque novedoso en el estudio de la base de datos aquí empleada. Ello, unido a la particularidad de la angioplastia practicada, brinda una relevancia especial a la investigación. Por otra parte, se ha observado que muchos de los índices de repolarización y despolarización están muy relacionados entre sí y por tanto aportan poco valor diferente al de otros que han resultado ser más significativos para identificar pacientes con isquemia en riesgo de sufrir MI. Finalmente, la cuidadosa metodología seguida para crear los diferentes modelos de predicción ha permitido que el modelo construido con las medidas más importantes tenga una mejor capacidad de generalización que la de otros modelos desarrollados previamente. El modelo puede servir como complemento a otros métodos estándar de diagnóstico de este tipo enfermedades.
Álvarez-García, Jesus; Vives, M.; Gomis, P.; Ordoñez, J.; Ferrero-Gregori, A.; Serra, A.; Cinca, J. Circulation Vol. 134, num. 18, p. e401-e402 DOI: 10.1161/CIRCULATIONAHA.116.024875 Data de publicació: 2016-11-01 Article en revista
This study aimed to develop a nonlinear support vector machine (SVM) model to detect ischemic events based on a dataset of QRS-derived and ST indices from nonischemic and acute ischemic episodes. The study included 67 patients undergoing elective percutaneous coronary intervention (PCI) with 12-lead continuous and signal-averaged ECG recordings before and during PCI. Fifty-four indices were initially considered from each episode. The dataset was randomly divided into training (80%) and testing (20%) subsets. The training subset was used to optimize the SVM parameters algorithm and for determining the most important statistically significant indices, by using repeated k-fold cross-validation (with N=25 repetitions and k=5). The described procedure was run on 25 randomized training/testing subsets to assess the average performance. On average, the most important indices were the QRSvector difference and the ST segment level at J-point + 60 ms computed from the synthesized vector magnitude, and the summed high-frequency QRS components of all 12 leads at 150 – 250 Hz band. The performance of testing was: classification error = 12.5(8.3 - 16.7)%, sensibility = 83.3(75.0 - 91.7)%, specificity = 91.7(83.3 - 91.7)%, positive predictive value = 90.9(83.0 - 92.3)% and negative predictive value = 85.7(80.0 - 91.7)%. The method used to construct the SVM model is robust enough and looks promising in detecting acute myocardial ischemia and myocardial infarction risk.
Calvo, M.; Le Rolle, V.; Romero, D.; Béhar, N.; Gomis, P.; Mabo, P.; Hernández, A. Annual International Conference of the IEEE Engineering in Medicine and Biology Society p. 5591-5594 DOI: 10.1109/EMBC.2016.7591994 Data de presentació: 2016-08 Presentació treball a congrés
This paper describes the integration of mathematical models of the cardiac electrical activity, the cardiovascular system and the baroreceptor reflex control of the autonomic nervous system, with a model representing a head-up tilt test. Sensitivity analyses are performed in order to determine those model parameters producing the greatest significant effects on heart rate and blood pressure. An optimization step is then applied to the most influential parameters in order to find the best model fit to real cardiac data obtained from a patient suffering from Brugada syndrome and a healthy subject, in supine and upright postures during a tilt test. Results show the feasibility of the model to generate realistic autonomic responses to head-up tilt testing in health and disease.
Álvarez-García, Jesus; Vives, M.; Gomis, P.; Ordoñez, J.; Ferrero-Gregori, A.; Serra, A.; Cinca, J. Circulation Vol. 133, num. 23, p. 2235-2242 DOI: 10.1161/CIRCULATIONAHA.116.021700 Data de publicació: 2016-06-07 Article en revista
Background-The arrhythmogenesis of ventricular myocardial ischemia has been extensively studied, but models of atrial ischemia in humans are lacking. This study aimed at describing the electrophysiological alterations induced by acute atrial ischemia secondary to atrial coronary branch occlusion during elective coronary angioplasty.; Methods and Results-Clinical data, 12-lead ECG, 12-hour Holter recordings, coronary angiography, and serial plasma levels of high-sensitivity troponin T and midregional proatrial natriuretic peptide were prospectively analyzed in 109 patients undergoing elective angioplasty of right or circumflex coronary arteries. Atrial coronary branches were identified and after the procedure patients were allocated into two groups: atrial branch occlusion (ABO, n= 17) and atrial branch patency (non-ABO, n= 92). In comparison with the non-ABO, patients with ABO showed: (1) higher incidence of periprocedural myocardial infarction (20% versus 53%, P= 0.01); (2) more frequent intra-atrial conduction delay (19% versus 46%, P= 0.03); (3) more marked PR segment deviation in the Holter recordings; and (4) higher incidence of atrial tachycardia (15% versus 41%, P= 0.02) and atrial fibrillation (0% versus 12%, P= 0.03). After adjustment by a propensity score, ABO was an independent predictor of periprocedural infarction (odds ratio, 3.4; 95% confidence interval, 1.01-11.6, P< 0.05) and atrial arrhythmias (odds ratio, 5.1; 95% confidence interval, 1.2-20.5, P= 0.02).; Conclusions-Selective atrial coronary artery occlusion during elective percutaneous transluminal coronary angioplasty is associated with myocardial ischemic damage, atrial arrhythmias, and intra-atrial conduction delay. Our data suggest that atrial ischemic episodes might be considered as a potential cause of atrial fibrillation in patients with chronic coronary artery disease.
Brugada syndrome is an inherited disease characterized by a specific electrocardiographic pattern as well as an increased risk of cardiac arrhythmias and sudden death . There are currently no objective indicators to determine individual risk, so the eligibility for the implantation of implantable cardioverter defibrillators to prevent sudden death events still generates controversy.
Since the electrocardiogram (ECG) provides with noninvasively heart information, this work presents an automatic algorithm able to detect significant electrocardiographic parameters in the disease, such as the Brugada-like pattern burden on 24-hour continuous ECG monitoring. The algorithm integrates a previously developed and validated ECG delineator in order to locate the beginning, peak and ending of cardiac waves . Then, it classifies each beat between normal or Brugada based on the morphology of the detected waves. Likewise, the algorithm proposes an expansion focused on detecting ST-segment elevation, PR, QRS and QTc intervals duration, heart rate variability and premature ventricular contractions load, among others.
The detector validation was based on the assessment of the 24-hour 12-lead ECG signals of a high-risk Brugada patient from the Hospital Clínic de Barcelona, in Spain. The subject carried the Holter monitor SEER 12 (GE Healthcare Inc., Milwaukee, WI, USA) for a whole day, placing the electrodes in a regular Mason-Likar 12-lead ECG system configuration and recording at a sampling frequency of 128 Hz. The obtained sensitivity and specificity of a random sample of 600 beats across the 12 leads (50 beats/lead) were of 93.7% and 98.69%, respectively. Erroneous detections resulted from incorrect T-wave delineations in noisy segments.
Future work will be focused on testing the performed algorithm in an extensive sample of patients diagnosed with Brugada syndrome that, based on their cardiac events, will automatically determine the association of these ECG parameters with the individual risk of suffering sudden cardiac death.
Aim: Baroreflex sensitivity (BRS) is a valuable index to assess cardiovascular autonomic control and therefore to provide prognostic evaluation in many cardiac diseases. Although several methods have been developed to capture spontaneous BRS noninvasively, they are difficult to compare and often provide conflicting results. This study analyzed eight different BRS estimates obtained from a clinical series of patients suffering from Brugada syndrome, in order to quantify the agreement between measures.
Methods: Thirty-seven patients diagnosed with Brugada syndrome took part in a head-up tilt test while noninvasive blood pressure recordings were collected. The test was divided into three phases, leading to 111 BRS results with each method: i) supine rest for 10 minutes, ii) tilting with an inclination of 60° for 45 minutes or until the test was positive and iii) rest in supine position for 10 minutes. Eight BRS estimates were compared in all phases: (1) BRS+/+ and (2) BRS-/- from sequence analysis, (3) BRS-LF, (4) BRS-HF, (5) BRS-LHF, (6) BRS-TF1 and (7) BRS-TF2 from cross-spectral analysis and (8) BRS-SD.
Results: Measures resulting from the sequence method failed to provide results in over 30 recordings, whereas BRS-SD and BRS-TF2 gave results in all cases. All methods captured a decrease in response to tilting, but BRS+/+ and BRS-/- obtained higher BRS after the test with respect to baseline. Moreover, these estimates showed low agreement with the remaining methods based on intra-class correlation coefficient results (BRS+/+: r = 0.21 ± 0.12, BRS-/-: r = 0.27 ± 0.15). Assuming an age-associated reduction in BRS, which has been previously reported, the lowest Pearson’s correlation coefficients came from sequence analysis results (r = -0.28 ± 0.03).
Conclusion: Among the BRS estimates analyzed, those derived from sequence analysis showed a lower reliability when capturing baroreflex function in a clinical series of Brugada syndrome patients.
El síndrome de Brugada es una enfermedad hereditaria caracterizada por un patrón electrocardiográfico característico, así como por un aumento del riesgo de sufrir arritmias cardíacas y muerte súbita. No existen actualmente indicadores objetivos que permitan determinar el riesgo individual, por lo que la elegibilidad para la implantación de desfibriladores automáticos implantables con el fin de prevenir eventos de muerte súbita genera todavía gran controversia. Dado que el electrocardiograma (ECG) proporciona información cardíaca de forma no invasiva, se presenta un algoritmo automático capaz de detectar parámetros electrocardiográficos relevantes en la enfermedad. De este modo, futuros estudios que incluyan grandes series de pacientes permitirán determinar el valor pronóstico de las diferentes variables detectadas. El algoritmo integra funciones disponibles previamente desarrolladas para la lectura y delineación de señales ECG y propone una expansión centrada en la detección de parámetros como la carga de latidos con patrón de Brugada, elevación del segmento ST, duración del segmento ST y de los intervalos PR, QRS y QTc, la variabilidad del ritmo cardíaco (VRC), la alternancia de la onda T o la carga de extrasístoles ventriculares, entre otros. El código se testeó en registros ECG continuos de 12 derivaciones y 24 horas de duración de dos pacientes con diferentes grados de severidad y un caso control, obteniendo tendencias relacionadas con la VRC y la carga de latidos con patrón de Brugada. Asimismo, se analizaron diferentes colocaciones de los electrodos con el objetivo de concluir la configuración más sensible para la detección del patrón característico.
High frequency (HF) QRS fragmentation and very-low amplitude abnormal intra-QRS potential (uAIQP) analyses have been used to track ischemic changes during coronary artery occlusions. The aim of this study was to assess the relationship between these two techniques in detecting acute myocardial ischemia and the effects of a previous myocardial infarction (MI).
Fifty-six patients who underwent elective percutaneous coronary intervention (PCI) procedures were selected and classified into 2 groups according to the presence of prior healed MI (old-MI) (n=18) or not (no_MI) (n=38). Continuous ECG before and during the PCI were
recorded and signal-averaged. uAIQPs were obtained using a signal modelling approach. HFQRSRMS was obtained by band pass filtering the ECGs at 150 to 250 Hz. QRS-HFpower was estimated from a modeling power spectral technique. uAIQP and HF indices were
obtained from a baseline and an occlusion-PCI ECG episode.
uAIQP and HF values decreased (p<0.05) for each of the 12 leads at the PCI event respect to baseline in all patients and the no-MI group. Changes in uAIQP or HF did not separate the groups. uAIQP and QRS-HFpower values at baseline were lower in all leads, except V1-V2,
in the old-MI groups compared to no-MI (p<0.05). Pearson’s correlation showed moderate relationship among the indices in most of leads.
High-frequency QRS fragmentation indices could add diagnostic value to ST analysis for diagnosing ischemia when a baseline ECG information is available. Patients with old-MI presented lower uAIQP amplitudes compared to no-MI, however further studies are needed to elucidate the effects of old MI on very-low level fragmentation of the QRS.
Introduction: Low-level electrocardiographic changes from depolarization wavefront may accompany acute myocardial ischemia. The purpose of this study was to assess the changes of microvolt amplitude intra-QRS potentials induced by elective percutaneous coronary interventions (PCI).; Methods: Fifty-seven patients with balloon inflation periods ranging from 3.1 to 7.3 minutes (4.9 +/- 0.7 min) were studied. Nine leads continuous high-resolution ECG before and during PCI were recorded and signal-averaged. Abnormal intra-QRS at microvolt level (mu AIQP) were obtained using a signal modeling approach. mu AIQP, R-wave amplitude and QRS duration were measured in the processed ECG during baseline and PCI episodes.; Results: The mean mu AIQP amplitude significantly decreased for each of the standard 12 leads at the PCI event respect to baseline. Left anterior descending artery (LAD) occlusion resulted in a decrease mu AIQP in both the precordial leads and the limb leads, while right coronary (RCA) and left circumflex (LCx) arteries occlusions mainly affected limb leads. R-wave amplitude increased during PCI in RCA and LCx groups in lead III but decreased in the precordial leads, while the amplitude decreased in the LAD group in lead III. The average duration of the QRS augmented in groups RCA and LCx but not in the LAD group.; Conclusions: Abnormal intra-QRS potentials at the level of mu V provide an excellent tool to characterize the very-low amplitude fragmentation of the QRS complex and its changes due to ischemic injuries. mu AIQP shows promise as a new ECG index to measure electrophysiologic changes associated with acute myocardial ischemia. (c) 2014 The Authors. Published by Elsevier Inc.
Myocardial ischemia may lead to significant changes in autonomic control of heart rate (HR) affecting its variability and alter beat-to-beat ventricular repolarization periods. We hypothesized that transient myocardial ischemia affects the complex dynamics of the HR and QT. The aim of this study was to assess the RR and QT interval time series complexity using Higuchi’s fractal dimension (HFD) during prolonged balloon occlusion of one of the major coronary arteries.
Eighty-five patients who underwent elective percutaneous coronary intervention procedures were selected. Patients were classified into 2 groups according to the presence of prior healed myocardial infarction (INF) (n = 29) or not (No_INF) (n = 56). RR, QT and QTc (Bazzet’s formula) time series were obtained from continuous ECG. Three 3-minute stages were chosen: (1) preinflation as baseline (pre), (2) from the start of occlusion (pci), and (3) immediately post deflation (post_pci). HFD was calculated on each 3-minute stage time series to quantify the changing complexity and self-similarity of RR, QT and QTc time series (HFD RR, HFD QT and HFD QTc, respectively). Cohen’s d statistics were included to measure the effect size of the procedure.
HFD values decreased significantly from baseline to pci and post_pci in the three time-series. HFD RR, 1.76 ± 0.13 vs. 1.69 ± 0.15 (p<0.001, Cohen’s d = -0.64); HFD QT, 1.90 ± 0.11 vs 1.82 ± 0.13 (p<0.001, d = -0.67); HFD QTc, 1.88 ± 0.09 vs. 1.83 ± 0.11 (p<0.01, d= -0.46). The HFD reduction was greater from baseline to post_pci (p<0.001, d = -0.99 (RR), d = -1.02 (QT)). HFD of QT intervals decreases during the procedure predominantly in the No_INF group.
The time series studied, related to autonomic control of HR and to the variability of ventricular repolarization, exhibit a reduced complexity provoked by transient myocardial ischemia.
Myocardial ischemia may lead to significant changes in autonomic control of heart rate (HR) affecting its variability and alter beat-to-beat ventricular repolarization periods. We hypothesized that transient myocardial ischemia affects the complex dynamics of the HR and QT. The aim of this study was to assess the RR and QT interval time series complexity using Higuchi's fractal dimension (HFD) during prolonged balloon occlusion of one of the major coronary arteries.
Autonomic nervous system (ANS) is governed by complex interactions arising from feedback loops of nonlinear systems that operate over a wide range of temporal and spatial scales, enabling the organism to adapt to stress, metabolic changes and diseases. This study is aimed to assess multifractal and nonlinear characteristics of the ANS during ischemic events provoked by a prolonged percutaneous coronary intervention (PCI) procedure. Eighty-seven patients from the STAFF III database were used. Patients were classified into 2 groups: (1) with prior myocardial infarction (MI) and (2) without MI (noMI). R–R signals during three 3-min stages of the procedures were analyzed using multifractal and surrogate data techniques. Multifractal indices increased significantly from the pre-inflation stage to the post-deflation stage. These variations were more marked for the noMI group. Multifractal changes significantly correlated with both the decreased parasympathetic and the increased sympathetic modulations accounted by classical linear indices. Multifractal measures resulted to be a more powerful indicator than linear HRV indices in quantifying the ischemia-induced changes. Right coronary artery (RCA) occlusions provoke greater multifractal reactions throughout the PCI procedure. Our findings suggest reduced complex multifractal and nonlinear reactions of ANS activity in patients with prior MI in comparison to the noMI group, possibly due to degradation in the complexity of control mechanism of heart rate generation
Cuponne, A.; Vallverdu, M.; Gomis, P.; Porta, A.; Voss, A.; de Luna, A. B.; Caminal, P. Computing in Cardiology Annual Conference p. 617-620 Data de presentació: 2012-09-11 Presentació treball a congrés
Repolarization dynamics may be of increasing interest in analyzing ECG-Holter for characterization of myocardial ischemic events related to cardiac death. The quantification of the dynamics of the beat-to-beat QT interval fluctuations, representing changes in repolarization duration, may be another emerging marker of cardiac events. Based on these arguments, we propose a symbolic analysis series to quantify the dynamics of the beat-to-beat QT interval fluctuations, representing changes in repolarization duration, and the prevalence of sympathetic or parasympathetic cardiac modulation in the RR series. This analysis decomposes the series in patterns of length L=3 beat and classify them into three categories: non-variable, variable, and very variable patterns referred to as P0, P1 and P2 patterns. The present work analyses QT and RR series obtained from 24-hour ECG-Holter recordings in order to obtained patterns able to stratify high (HRG) and low risk (LRG) of suffer cardiac mortality in patients with symptomatic myocardial ischemia. Comparing LRG and HRG, results showed that pattern P0 could better quantify QT series and pattern P2 the RR series. These findings suggest a decreased cardiac vagal function with a relative increase in sympathetic cardiac modulation, and more complex pattern of ventricular repolarization in the HRG.
Gomis, P.; Caminal, P.; Vallverdu, M.; Warren, S.; Stein, P.; Wagner, G. Journal of electrocardiology Vol. 45, num. 1, p. 82-89 DOI: 10.1016/j.jelectrocard.2011.08.013 Data de publicació: 2012-01 Article en revista
Magrans, R.; Gomis, P.; Voss, A.; Caminal, P. Annual International Conference of the IEEE Engineering in Medicine and Biology Society p. 7083-7086 Data de presentació: 2011-09-03 Presentació treball a congrés
The purpose of this study was to describe the
changes in high-frequency QRS (HF-QRS) components due to
myocardial ischemia provoked by prolonged artery occlusion
during percutaneous coronary interventions (PCI). Signalaveraged
ECGs from 69 patients were obtained during PCI
procedures and comparison of high-frequency components of
the QRS at different temporal regions and frequency
bandwidth were performed. Continuous wavelet transform was
applied to estimate the energy contents over the studied timefrequency
regions. Seven frequency bands from 50 to 300 Hz,
with bandwidth = 100 Hz were considered. The sum of all 12
leads energy decreased significantly (p < 0.001) from pre-PCI
to PCI during both second half and total QRS complex in all
frequency bands, but the main effect was found in the 200-300
Hz band. The energy changes were more marked toward
higher frequency bands. The second half of QRS complex was
more sensitive to changes due to myocardial ischemia.
A new method for myocardial ischemia detection is proposed in this communication. The originality of this method relies on the
consideration of the dynamics of times series extracted from the ECG whereas traditionnal methods are simply based on static measures. After the extraction of a feature vector, the dynamics are caracterised with an Hidden Semi-Markovian Model (HSMM). The ischemic detector uses
a reference HSMM and an ischemic HSMM and then compare the likelihood of the time series. Results obtained with PTCA (percutaneous transluminal coronary angioplasty) records of the STAFF3 database show a very good detection rate (96% of sensibility and 80% of specificity).
Magrans, R.; Gomis, P.; Caminal, P.; Wagner, G. Annual International Conference of the IEEE Engineering in Medicine and Biology Society p. 1808-1811 DOI: 10.1109/IEMBS.2009.5333930 Data de presentació: 2009-09-03 Presentació treball a congrés
We assess the autonomic nervous system response during prolonged percutaneous transluminal coronary angioplasty (PTCA) by heart rate variability analysis using
multifractal indices. These indices are also used to evaluate the effects of the PTCA at different arteries. The indices augmented significantly during transient ischemia and
reperfusion periods indicating an increase of multifractal degree and a decrease of the long-range dependence on heartbeat fluctuations. This indicates that significant
multifractal complex reactions of autonomic control of the heart rate occurred during coronary artery occlusions.
Benitez, R.; Alvarez-Lacalle, E.; Echebarria, B.; Gomis, P.; Vallverdu, M.; Caminal, P. Medical engineering and physics Vol. 31, num. 6, p. 660-667 DOI: 10.1016/j.medengphy.2008.12.006 Data de publicació: 2009-07 Article en revista
Ng, F.; Wong, S.; Gomis, P.; Lim, J.; Passariello, G.; Ansermino, J. Annual International Conference of the IEEE Engineering in Medicine and Biology Society p. 4692-4695 Data de presentació: 2008-08-22 Presentació treball a congrés