HeartSafe – Assessing Heart Function for Unsupervised Homecare Applications through Multi-Channel Auscultation

Description

The main scientific goal of this project is to research algorithms applicable in long-term home monitoring contexts for the assessment of cardiac function, the key variable in cardiovascular disease (CVD) management, using a multi-channel heart sound auscultation approach. Clinical applications will be developed for early detection of Heart Failure (HF) decompensation episodes in elderly and neonatal populations; both will be clinically assessed using in-hospital studies. The project will build on top of the results achieved by the team in heart sound processing using single channel acquisition during the EU HeartCycle project (FP7-216695) and two projects financed by FCT, i.e., SoundForLife (PTDC-EIA-68620-2006) and DigiScope (PTDC/EIA-CCO/100844/2008). Commercial interest in the project’s outcomes has been expressed by Critical Software (see attached letter). CVD are the leading cause of death (45%) in developed and developing countries and one of the major causes of hospitalization. CVD are highly constraining diseases, tightly associated to aging. In Europe, the proportion of the population aged >65 years is projected to increase to 20% in 2025. These two factors are inducing major pressures on Europe’s health care systems. Heart Failure (HF) – a CVD being the most frequent cause of hospitalization for people over 65 – has 10 million patients in the EU. Daily (home) monitoring and close follow up is lacking. This leads to frequent decompensation episodes with hospitalization [1] and ultimately to patient’s death. At the other extreme of life, neonates often have prolonged intensive care stays due to cardiovascular decompensation caused by a persistent arterial duct or pulmonary hypertension of the newborn. This is another group of patients in whom the study of cardiovascular function poses special challenges. Effective preventive healthcare strategies based on home care approaches require robust and inexpensive solutions for early detection of HF decompensation episodes for long-term patient follow-up. In this context, heart sound is a very valuable vital sign, since it directly encodes the mechanical status of the heart. As such, it was key diagnosis tools in clinical practice up to the introduction of echocardiography. Heart sound processing will be applied to extract three different key diagnostic measures to characterize the cardiovascular system status and to prognosis HF: the systolic time intervals (STIs), S3 sounds and S2 split. Myocardial relaxation and contraction are governed by intracellular recycling of calcium ions. The timings of these basic cardiac events are directly related to the health of the cardiac cells [2]. Of major importance are the systolic time intervals (STI) of the left ventricle. Unhealthy hearts will exhibit noticeable deviations from normal STIs. This diagnosis tool was the gold standard prior to the introduction of echocardiography. The third heart sound (S3) is considered pathological in population over 40 years old. Its occurrence has been related by several clinical studies to high levels of NT-proBNP, which is an amino acid polypeptide segregated by the ventricles of the heart in response to excessive stretching of heart muscle cells and is a highly relevant diagnostic and prognostic marker for left ventricle HF [ ]. The second heart sound (S2) is mostly a mixture of two quasi-simultaneous sounds: the closing of the aortic (A2) and the pulmonary (P2) heart valves [15]. While A2 is a reasonably stable sound, P2 depends on the respiration cycle and is slightly delayed during inspiration, creating a window of opportunity in which these sounds are split and can be heard individually. P2 is typically hyper-phonetic in individuals with pulmonary hypertension, making us believe that it might be possible to use signal processing to segment these heart sounds and possibly detect pulmonary hypertension situations. Innovation Heart sound has proven to be a highly informative diagnostic tool to assess the status of the cardiovascular state of a patient. However, since it is very prone to noise interference, its application in clinical practice requires highly controlled acquisition environments. Using the multi-channel setup proposed in this project, signal redundancy will be explored for robust noise cancelling and blind source separation techniques in order to achieve a signal with diagnostic value even under uncontrolled conditions. Furthermore, the signal redundancy will also enable more accurate detection schemes of the most pertinent diagnostic measures, i.e., STIs, S2-split and S3.

Researchers

Funded by

FCT

Partners

IT - Porto, CHUC

Total budget

200 000,00 €

Keywords

Heart Sound, Biosignal Processing, pHealth, Pattern Recognition

Start Date

2013-04-01

End Date

2016-03-31

Journal Articles

Conference Articles