Morphologic clustering of earcanals using deep learning algorithm to design artificial ears dedicated to earplug...

Area of Science:

• Audiology
• Biomedical Engineering
• Acoustics

Background:

• Designing effective earplugs requires acoustical test fixtures (ATFs) that accurately represent diverse human ear canal morphologies.
• Existing ATFs often use simplified, straight cylindrical ear canals, limiting their ability to assess real-world earplug performance across different users.

Purpose of the Study:

• To develop a methodology for clustering ear canal morphologies to inform the design of artificial ears for sound attenuation measurement.
• To identify key ear canal morphologic indicators influencing the attenuation performance of commercial earplugs.

Main Methods:

• A sample of Canadian workers' ear canals was analyzed.
• Statistical analysis and an artificial intelligence-based algorithm were used to cluster ear canal morphologies.
• Correlation between ear canal features and earplug sound attenuation was investigated.

Main Results:

• Three distinct ear canal clusters were identified based on morphology.
• Key differentiating features included ear canal length and the surface and ovality of the first bend's cross-section.
• Earplugs demonstrated significantly higher attenuation in ear canals with smaller girth and rounder first bends compared to larger, more oval ones.

Conclusions:

• Ear canal morphology significantly impacts earplug sound attenuation.
• The developed clustering methodology provides a basis for designing more representative artificial ears for ATF development.
• This research highlights the need for earplug testing that accounts for individual ear canal variations.