Cortical benefits of DNN-based noise management in experienced hearing aid users measured using functional near infrared spectroscopy (fNIRS)

There is growing interest in understanding how emerging hearing aid technology affects listening effort, particularly in noisy environments. This study used functional near-infrared spectroscopy (fNIRS) to examine the impact of a deep neural network (DNN)-based noise management program on cortical oxygenation in brain areas associated with listening effort. We hypothesized that a DNN-based program would be associated with less cortical oxygenation in the left prefrontal cortex compared to a traditional quiet-listening hearing aid program. Twenty-six experienced hearing aid users listened to sentences in noise and repeated the sentence-final words while using two hearing aid programs: a standard quiet-listening program with omnidirectional microphones, and a noise management program combining directional microphones with the DNN-based denoising algorithm. Outcome variables included listening accuracy, subjective listening effort, and oxygenation in the prefrontal cortex. The noise management program led to significantly better performance—participants showed higher accuracy, reported lower effort, and exhibited reduced activation in the left prefrontal cortex. Notably, the size of this benefit was comparable to that seen when comparing aided to unaided listening in a similar acoustic scene. These findings highlight the utility of fNIRS for assessing listening effort in hearing aid research and demonstrate the potential of deep learning-based hearing aid processing to ease cognitive demands during communication. Results offer valuable insight into future hearing aid development and evaluation.