Abstract— Systolic time intervals (STI) have significant diagnostic and prognostic value to assess the global cardiac function. Presently, STIs are regarded as a promising tool for long-term follow-up of patients with chronic cardio- vascular diseases. Heart sound has proven to be a valuable approach for STI estimation, in particular for the Pre- Ejection Period (PEP). However, since the optimal auscultation site varies from individual to individual, as well as with the position of the body, its application in single- channel and fixed auscultation site setups poses practical difficulties. Hence, we extend our previous work on PEP estimation to a multi-channel sound acquisition setup, where signal redundancy is exploited. A channel selection method is proposed and the best channel is selected for PEP estimation. As a preliminary study, the devised algorithms were evaluated with respect to echocardiography reference on a set of 236 heartbeats collected from 8 healthy subjects in two sound auscultation sites. The channel selection approach led to 8.4% estimation error decrease, in comparison to a single- channel approach. Current results support our assumption that a multi-channel audio-based strategy can be applied to assess STI in personal health application scenarios. I. INTRODUCTION The timings of myocardial contraction and relaxation are directly related to the health of cardiac cells [1] as they determine the ability of the myocardium to achieve blood delivery according to the metabolic requirements of the organs. The systolic and diastolic timings of the left ventricle assume particular relevance, since this ventricle is responsible for blood flow in the systemic circulation. Clinically accepted descriptions of systolic function of the left ventricle are the velocity of pressure rise, the velocity of ejection, the extent of ejection and the ejection fraction [2]. These function indicators can be obtained using both invasive and non-invasive procedures (e.g. echocardiography, which is the current gold standard for systolic time intervals measurement) in in-hospital settings. However, these procedures are not adequate for daily applications in home settings, as required for long-term follow-up of patients with chronic cardiovascular diseases. An adequate alternative to evaluate the global cardiac function in this type of application scenarios is the use of systolic time intervals (STI). In fact, several studies have This work was partly supported by the HeartSafe project (PTDC/EEI- PRO/2857/2012) financed by the Portuguese Foundation for Science and Technology and the iCIS project (CENTRO-07-0224-FEDER-002003). R. P. Paiva, T. Sapata, P. Carvalho and J. Henriques are with CISUC, Department of Informatics Engineering, University of Coimbra, Portugal (e-mail: {ruipedro, sapata, carvalho, jh} @dei.uc.pt). I. Quintal, R. Baptista and L. Gonçalves are with the Echocardiography Department of the Centro Hospitalar e Universitário de Coimbra, Coimbra, Portugal (e-mail: isabelquintal@chc.minsaude.pt). shown that STIs are highly correlated to major and fundamental cardiac functions [3-6]. Systolic ejection is preceded by the electro-mechanical delay and by the isovolumetric contraction time. These two time intervals compose the pre-ejection period (PEP), which is the time interval between the start of ventricular depolarization and the moment of aortic valve opening. The left ventricle ejection time (LVET) is defined as the time interval of left ventricular ejection, which occurs between the opening of the aortic valve and its subsequent closure. PEP and LVET assume major relevance in assessing the cardiac reserve and the left ventricular function [7-8]. PEP is an index of the left ventricular function and reflects changes in myocardial contractility, left ventricular end-diastolic volume and aortic diastolic pressure. Another important application of PEP is in non-invasive beat-by-beat estimation of blood pressure [8]. The left ventricular ejection period can also be related to contractility and to cardiac output [9]. It is by itself a measure of cardiac function. Several measurement modalities for STI assessment in home settings or other scenarios where portability is recommended have been considered in the literature. Namely, photoplethysmography [10], radial pulse pressure [11], phonocardiography [12] and impedance cardiography [13] are subject of current research efforts. Heart sound (HS) has proven to be a highly informative diagnostic tool to assess the status of the cardiovascular state of a patient. In fact, in the past we introduced a phonocardiography-based STI measurement algorithm based on a Bayes approach [12], which has shown to exhibit the highest accuracy, precision and correlation with respect to the gold standard among available methods for portable STI measurement [14]. However, the optimal auscultation site varies from individual to individual, as well as with the position of the body. Therefore, single-channel, fixed auscultation site setups, like the one we employed in our past work, pose practical difficulties. On a controlled environment, a single- channel approach could be sufficient in case a specialist would look for and select the best auscultation site. However, this is not feasible for home-monitoring scenarios, where, typically, sound acquisition would be acquired from one or more fixed sites (e.g., sensors integrated into a vest). To evaluate the potential of this idea, we propose an extension of our previous work on PEP estimation to a multi-channel, fixed sites, sound acquisition setup. Signal redundancy is exploited and the channel with the highest quality is selected, thus allowing for better PEP estimation accuracy. In our experiments, a preliminary two-channel approach achieves 8.4% higher accuracy than the single- channel strategy. Multi-Channel Audio-based Estimation of the Pre-Ejection Period R. P. Paiva, T. Sapata, J. Henriques, I. Quintal, R. Baptista, L. Gonçalves, P. Carvalho CONFIDENTIAL. Limited circulation. For review only. Preprint submitted to 37th Annual International Conference of the IEEE Engineering in Medicine and Biology Society. Received April 10, 2015.