Performance evaluation of implantable artificial organs by sound spectrum analysis.

In this paper, a sound spectrum analyzing method was proposed to pre detect malfunctions of implantable artificial organs, such as an electromechanical total artificial heart (TAH) or prosthetic valves, without any percutaneous invasion. For this purpose, a sound detecting device was developed using a high sensitivity condenser microphone with a frequency range of more than 13 kHz. Output signals of this device are sampled at 100 kHz maximally, and sampled data are stored in an IBM PC (SamBo, Korea). To remove environmental noises in the measured sound, an adaptive least-mean square algorithm was employed. Using the squared value of the sound signal, the best position where only sounds from mechanical components can be measured was found. The sound spectrum was obtained by the periodogram spectral estimating method. Experiments were performed with this system, and the results indicated that: 1) by using an adaptive noise cancelling algorithm, a more noise-free signal can be obtained; 2) the harmonics from the mechanical components of a pendulum type electromechanical TAH were approximately 1.3 KHz; 3) a spectral change was observed when we compared the power spectral densities of a normal and failed TAH; 4) the spectral shift to higher harmonics occurred with an increase in heart rate; and 5) the sound propagation properties of tissue were investigated with animal experiments. The method proposed was found to be applicable in the detection of implantable artificial organ mechanical failure by sound spectrum analysis without the need for percutaneous invasion.