J.S. Fleeter -jen@fc.hp.com
G.R. Wodicka
Purdue University,West Lafayette, IN 47907
Popular version of paper 2pSPb4
Presented Tuesday afternoon, October 13, 1998
136th ASA Meeting, Norfolk, VA
Auscultation is defined as the process of listening for sounds that come from the human body, usually with the aid of a stethoscope. Many diagnoses concerning the health of both the respiratory and cardiac systems are made using a standard binaural stethoscope. However, the stethoscope is extremely sensitive to interference from ambient noise. In high noise environments, the standard stethoscope is essentially useless for diagnostic purposes.
Ambulances, helicopters, and fixed wing aircraft are typically used to evacuate patients in an emergency. Each of these vehicles can be considered to be a high noise environment where a standard stethoscope is not usable. For example, the noise level inside the cabin of the C-130 aircraft used by the US Air Force has ambient noise levels of 90 to 100 dB. Noise levels of 90 120 dB exist inside a helicopter. In contrast, in a healthy adult, normal body sounds are in the range of 22-30 dB in free space, and 65 to 70 dB through a stethoscope. A traditional stethoscope is not an effective medical tool in these environments.
An active noise reduction (ANR) stethoscope and digital signal processing techniques were used to extract heart and lung sounds from background noise. Noise corrupted heart and lung sounds were collected from a subject using a two-microphone coupler in a simulated aircraft environment. Several experiments were run to determine the best combination of stethoscope coupler geometry and digital signal processing algorithm for the recovery of heart and lung sounds in high noise environments.
The performance of two common signal processing algorithms were tested for this application. When the first coupler was designed, the Least Mean Squared (LMS) algorithm was used to process the coupler microphone signals. In a previous study, it was shown in off-line calculations that the more complicated Normalized Least Mean Squared (NLMS) algorithm would produce a recovered signal that would be easier to hear. Therefore, this algorithm was put into our real-time system.
Two stethoscope couplers were also compared. The stethoscope couplers tested were extremely similar to one another. Both use primary and reference microphones. The primary signal is from a microphone that points towards the subject and measures that sounds emanating from the body. The reference signal is measured by a microphone that points away from the subject, and measures surrounding noise. The original coupler design had a chamber for the primary microphone, much as one would find in a classic stethoscope. However, the reference microphone, was open to the surrounding air only via a small cylindrical port. This, it was considered that because of the different shapes of chambers leading to the two microphones, that the reference noise signal may not be a very accurate representation of the actual noise we were trying to cancel. Therefore, the shape of the port leading to the reference microphone was changed so that it was more similar to the one leading to the primary microphone.
Lastly, the noise source was changed. A majority of the preliminary work on this stethoscope system was done assuming that it would be used inside a fixed-wing aircraft such as a C-130. However, if this is not the case, and it is used inside other vehicles, such as a helicopter, it might be more difficult to recover heart and lung sounds.