Brain benefits of deep learning-based noise management in experienced hearing aid users using functional near infrared spectroscopy

Abstract

Abstract There is growing interest in using neuroimaging to understanding listening effort in individuals with hearing loss, with a particular focus on how innovative hearing aid features impact listening effort. This study used functional near infrared spectroscopy (fNIRS) to investigate the neural basis of listening effort in experienced hearing aid users. This study evaluated the impact of a new deep learning-based noise management hearing aid feature on cerebral blood oxygenation, with the expectation that it would be associated with less oxygenation in the left prefrontal cortex compared to a standard quiet-listening hearing aid program. Twenty-six experienced hearing aid users repeated sentence-final words from sentences presented in noise while wearing individually prescribed hearing aids in two conditions: a standard-listening program with an omnidirectional microphone setting and a DNN-listening program combining directional microphones with a deep-neural-network-based noise management algorithm. Listening accuracy, subjective listening effort ratings, and prefrontal oxygenation via fNIRS were measured throughout testing. As expected, the DNN-listening program was associated with higher listening accuracy, lower subjective listening effort ratings, and lower fNIRS-measured oxygenation in the left prefrontal cortex relative to the standard-listening program. The utility of fNIRS for hearing aid research and the interaction of listening effort and other cognitive processes are discussed further.