An open access database for the evaluation of heart sound algorithms

International audienceIn the past few decades, analysis of heart sound signals (i.e., the phonocardiogram or PCG),especially for automated heart sound segmentation and classification, has been widely studied and hasbeen reported to have the potential value to detect pathology accurately in clinical applications. However,comparative analyses of algorithms in the literature have been hindered by the lack of high-quality,rigorously validated, and standardized open databases of heart sound recordings. This paper describes apublic heart sound database, assembled for an international competition, the PhysioNet/Computing inCardiology (CinC) Challenge 2016. The archive comprises nine different heart sound databases sourcedfrom multiple research groups around the world. It includes 2,435 heart sound recordings in totalcollected from 1,297 healthy subjects and patients with a variety of conditions, including heart valvedisease and coronary artery disease. The recordings were collected from a variety of clinical ornonclinical (such as in-home visits) environments and equipment. The length of recording varied fromseveral seconds to several minutes. This article reports detailed information about the subjects/patientsincluding demographics (number, age, gender), recordings (number, location, state and time length),associated synchronously recorded signals, sampling frequency and sensor type used. We also provide abrief summary of the commonly used heart sound segmentation and classification methods, includingopen source code provided concurrently for the Challenge. A description of the PhysioNet/CinCChallenge 2016, including the main aims, the training and test sets, the hand corrected annotations fordifferent heart sound states, the scoring mechanism, and associated open source code are provided. Inaddition, several potential benefits from the public heart sound database are discussed