0
2010 Design of Medical Devices Conference Abstracts

Artificial Neural Network Analysis of Heart Sounds Captured From an Acoustic Stethoscope and Emailed Using iStethoscopePro OPEN ACCESS

[+] Author and Article Information
Dustin Palm

Medical School, University of Minnesota

Stan Burns

University of Minneosta Duluth

Trichy Pasupathy, Eric Deip, Brittney Blair, Misty Flynn, Amanda Drewek, Matt Sjostrand, Brian Stephenson

Itasca Community College

Glenn Nordehn

University of Minnesota Duluth

J. Med. Devices 4(2), 027531 (Aug 11, 2010) (1 page) doi:10.1115/1.3443737 History: Published August 11, 2010

Abstract

Valvular heart disease is a significant problem. The primary care physician initially does assessment through auscultation. Accuracy in classification of sounds is suboptimal (20–40%). Technological advances have paralleled an increase in referral for Doppler echocardiography and a decrease in auscultatory skill. An increase in the referral of functionally innocent heart murmurs has contributed to the increasing cost of care. A computer-aided analysis has been shown to improve the accuracy of primary care physicians. A remote centralized computer-aided analysis could provide physicians with an additional tool in the assessment of heart murmurs, especially in settings without access to echocardiography. iStethoscopePro is an application for the iPhone and iPod Touch capable of recording and emailing sounds. We developed a device, which interfaces with iStethoscopePro and any acoustic stethoscope. We used this device to capture heart sounds from a conventional acoustic stethoscope and email them using iStethoscopePro for analysis with an artificial neural network (ANN). Hypothesis: It is possible to record heart sounds from an acoustic stethoscope, email them, and classify them with an ANN. Our device recorded heart sounds with insignificant intersample variation. After training the ANN with representations of four heart murmurs (aortic regurgitation, aortic stenosis, mitral regurgitation, and mitral stenosis) and normal, we achieved an overall accuracy of 45% with sensitivities of 50–75%. A remote centralized analysis of sound captured from an acoustic stethoscope is possible and could augment traditional auscultatory exams by offering an objective classification. Improving the accuracy and specificity of the ANN is necessary. This collection modality offers a method for the collection of a great deal of sounds for further development of artificial intelligence systems.

Copyright © 2010 by American Society of Mechanical Engineers
This article is only available in the PDF format.

References

Figures

Tables

Errata

Discussions

Some tools below are only available to our subscribers or users with an online account.

Related Content

Customize your page view by dragging and repositioning the boxes below.

Topic Collections

Sorry! You do not have access to this content. For assistance or to subscribe, please contact us:

  • TELEPHONE: 1-800-843-2763 (Toll-free in the USA)
  • EMAIL: asmedigitalcollection@asme.org
Sign In