Title: A Contact-less System for the Detection of Atrial Fibrillation using Camera Recordings
Short name: AFCam
People involved: Luca T. Mainardi, Filippo Molinari, Valentina Corino, Federico Lombardi
Funding source: Alta scuola politecnica
Funding period: 2016-2017
Partners: Cardiologia, Ospedale Maggiore Policlinico, Università Statale di Milano
Short description: Development of innovative, contactless system for AF diagnosis based on extraction of cardiopulmonary parameters from video recording of subject’s face.
Aim: Atrial fibrillation (AF) is the most common arrhythmia that affects 1–2% of the population. Paroxysmal AF (PAF) is a subtype of AF which is characterized by periods of episodically irregular rhythm, mostly asymptomatic. If undiagnosed, it leads to an increased risk of stroke and heart failure. Being largely asymptomatic, PAF is only occasionally detected by short time monitoring periods, leading to a bunk of untreated patients with high risk of AF complications and stroke. Therefore we propose to develop an innovative, contactless system for PAF diagnosis based on robust extraction of cardiopulmonary parameters from video recordings of subject’s face. The objectives of the project are:
1) definition of a product/solution for PAF diagnosis.
2) the definition of robust face tracking algorithm to recognize subject’s face and tracks its movements;
3) the extraction of reliable video-derived blood volume pulse (BVP) signal and development of accurate AF detector;
4) the validation of the system in clinical environment with the recording of 70 patients at Ospedale Maggiore Policlinico, Milano;
5) definition of use-case scenarios.
Expected results: We expect from the team:
a) The design and the requirements definition of the AFCam product.
b) Definition of a real-time robust method to process Video images and to extract VideoPPG signal enhancing from noise sources due to subject’s movements and ambient light changes;
c) Algorithm that can discriminate between three classes of subjects: AF patients, patients with normal sinus rhythm and patients affected by other forms of arrhythmia.
d) Recruitment of at least 70 subjects with different health state condition: AF, normal sinus rhythm, other arrhythmias
e) Identification of use-case scenarios in different environments (Hospitals, home, public places…)