Identification of hearing loss using TEOAE descriptors: theoretical foundations and preliminary results.

It was hypothesized that the relationship between transiently evoked otoacoustic emission (TEOAE) signals and the functional status of the outer hair cells provides an opportunity to design a clinical procedure that can evaluate the normality of cochlear function. To discriminate normal subjects from subjects suffering from mild to moderate hearing loss (HL), it was assumed that every subject population has unique and discrete TEOAE signal descriptors. The main classification algorithm was based on a discriminant analysis of raw fast Fourier transform data. When it was applied to a sample set of TEOAE recordings (from 56 normal and 68 HL subjects) elicited from 68-dB SPL click stimuli, it correctly identified 90.2% of the normal and 87.5% of the HL subjects. The same algorithm yielded an 85.5% discrimination between TEOAE recordings from conductive and cochlear HL cases.