A Non-intrusive, Non-obstructive, Versatile Venous Reservoir Blood Volume Sensor Based on Computer Vision for Clinical Cardiopulmonary Bypass

Abstract

We present an automatic venous reservoir blood volume sensor system, based on computer vision, for clinical cardiopulmonary bypass. A camera connected to a computer reads the blood volume inside the venous reservoir in real time, using custom-developed algorithms and software. The system's performance in blood volume measurement has been characterized, achieving a mean absolute percentage error as small as 2.5%. Through comprehensive testing in a simulated clinical environment, the system has proven to be nonintrusive to blood, nonobstructive for operators, and versatile for different reservoir models.