A Deep Neural-Network Classifier for Photograph-Based Estimation of Hearing Protection Attenuation and Fit

Details

• Alternative Title:
  J Acoust Soc Am
• Personal Author:
  Smalt, Christoper J. ; Ciccarelli, Gregory A. ; Rodriguez, Aaron ; Murphy, William J.
• Description:
  Occupational and recreational acoustic noise exposure is known to cause permanent hearing damage and reduced quality of life, which indicates the importance of noise controls including hearing protection devices (HPDs) in situations where high noise levels exist. While HPDs can provide adequate protection for many noise exposures, it is often a challenge to properly train HPD users and maintain compliance with usage guidelines. HPD fit-testing systems are commercially available to ensure proper attenuation is achieved, but they often require specific facilities designed for hearing testing (e.g., a quiet room or an audiometric booth) or special equipment (e.g., modified HPDs designed specifically for fit testing). In this study, we explored using visual information from a photograph of an HPD inserted into the ear to estimate hearing protector attenuation. Our dataset consists of 960 unique photographs from four types of hearing protectors across 160 individuals. We achieved 73% classification accuracy in predicting if the fit was greater or less than the median measured attenuation (29 dB at 1 kHz) using a deep neural network. Ultimately, the fit-test technique developed in this research could be used for training as well as for automated compliance monitoring in noisy environments to prevent hearing loss.
• Subjects:
  Article Ear Protective Devices Hearing Hearing Loss, Noise-Induced Humans Neural Networks, Computer Noise, Occupational Quality Of Life
• Source:
  J Acoust Soc Am. 150(2):1067
• Pubmed ID:
  34470332
• Pubmed Central ID:
  PMC8689361
• Document Type:
  Journal Article
• Funding:
  CC999999/ImCDC/Intramural CDC HHSUnited States/
• Volume:
  150
• Issue:
  2
• Collection(s):
  CDC Public Access
• Main Document Checksum:
  urn:sha256:2d940ad379def9b83bba760004f9f595097d2a817b386476623bc7e0ecec4a0f
• Download URL:
  https://stacks.cdc.gov/view/cdc/112819/cdc_112819_DS1.pdf
• File Type:
  [PDF - 2.00 MB ]