Remote heart sound characterisation and classification using computational imaging

Cardiac dysfunction is a leading cause of death that might be prevented by early diagnosis and timely treatments. Auscultation of the heart with a stethoscope is a key component of clinical examination. In a healthy individual, for each heartbeat, two heart sounds (S1 and S2) can be heard, produced by the closure of cardiac valves. Additional heart sounds (for example S3 and S4) might indicate a severe cardiac damage and a poor prognosis, but their identification is difficult, requires training, and depends on clinical experience [1] [2] . Therefore, easily deployable technologies that allow contactless and non-subjective detection of heart sounds are desirable.