The I. A. and Fannie R. Barnett Lecture Series

Wideband acoustic immittance (WAI)—also known as wideband reflectance—offers a powerful, noninvasive method for detecting a range of middle-ear disorders using absorbance and impedance measurements taken in the ear canal in a certain frequencies range. Despite its diagnostic potential, WAI remains challenging in clinical practice due to factors that can corrupt measurements, such as air leaks, calibration errors, or poorly positioned probes. Among these, undetected air leaks are particularly problematic, as they can lead to misinterpretation and unnecessary referrals.

In this talk, I will present recent joint work with the Voss Lab at Smith College, where we developed machine learning (ML) methods to automatically detect acoustic leaks in ambient-pressure WAI measurements. The project has two main goals: (1) to explore model paradigms for applying ML techniques to WAI data, and (2) to offer a practical tool that alerts clinicians in real time when a leak is present, allowing the measurement to be repeated immediately and improving diagnostic reliability.

Refreshments will be provided in the Math Department Faculty Lounce (4118 French Hall West) from 3:15-3:45pm preceding the lecture.