UC Irvine

Objectives

To assess the validity of a bench-top model of an optical tympanometry device to diagnose in vitro model of middle ear effusion (MEE).

Methods and materials

We illuminated an in vitro model of ear canal and tympanic membrane with broadband light and relayed remitted light to a spectrometer system. We then used our proprietary algorithm to extract spectral features that, together with our logistic regression classifiers, led us to calculate a set of simplified indices related to different middle ear states. Our model included a glass vial covered with a porcine submucosa (representing the tympanic membrane) and filled with air, water, or milk solution (representing different MEE), and a set of cover-glass slips filled with either blood (representing erythema) or cerumen. By interchanging fluid types and cover-glass slips, we made measurements on combinations corresponding to normal healthy ear and purulent or serous MEE.

Results

Each simulated condition had a distinct spectral profile, which was then employed by our algorithm to discriminate clean and cerumen-covered purulent and serous MEE. Two logistic purulent and serous MEE classifiers correctly classified all in vitro middle ear states with 100% accuracy assessed by leave-one-out and k-fold cross validation.

Conclusions

This proof-of-concept in vitro study addressed an unmet need by introducing a device that easily and accurately can assess middle ear effusion. Future in vivo studies aimed at collecting data from clinical settings are warranted to further elucidate the validity of the technology in diagnosing pediatric acute otitis media.