ACOUSTIC SENSING TO DETERMINE THE CONDITION OF IMPLANTED BLOOD PUMPS

Abstract

Both surgeon and patient need to know the running condition of blood pumps. Undetected faults can result in emergency surgery or loss of life, while replacement based on elapsed time exposes the patient to unnecessary surgical risk and increases costs. Condition monitoring can provide early detection of faults in ventricular assist systems (VAS) and total artificial hearts (TAHs), by measuring vibration signals characteristic of component degradation. The effectiveness of this approach depends on sensitive detection of critical signal frequencies, while they are still at low amplitude values. This project compared monitoring of fault-related vibration signals by implanted sensors on the pump to an acoustic sensor applied at the skin. Month-long animal tests were conducted using implanted generators, excited by vibration frequencies and amplitudes typical of rotary blood pumps. Given proper attention to sensor selection, sensor placement, and data processing, the results showed that frequencies of interest can be detected acoustically and separated from physiologic and background sounds. Side-by-side comparisons of an acoustic sensor at the skin to accelerometers on the implanted generators showed an ability to detect low levels of important frequencies through 5+ cm of bone, muscle, and subcutaneous fat tissue. These results indicate that an effective, noninvasive condition monitor for blood pumps can be developed using proven technologies.