Neural perceptions of auditory stimuli for the assessment and measurement of hearing acuity

Pure-tone Audiometry has remained fairly stagnant for many years. Pure-tone audiograms are a behavioural test, meaning the patient must be able to understand instructions and press a response button to participate. This test forms a critical part of Audiometry as a field, allowing for detailed and frequency-specific hearing thresholds to be extracted, a key tool for the calibration of hearing aids that are specifically calibrated to a user's hearing loss. For adults with physical or mental disabilities, the current Pure-Tone Audiogram cannot be undertaken, excluding these minority groups from accurate hearing assessments. Audiometry as a whole has progressed, with interesting and useful new methods such as the Auditory Brainstem Response changing how audiometry is performed in infants, with most infants undertaking one within the first months of their life. This test allows the detection of hearing loss in infants, but the response diminishes quickly with age making it unsuitable for use in adults. When it comes to obtaining frequency specific thresholds without a behavioural response, there are currently no available methods for adults, and the existing literature in this domain typically focuses on modulated tones designed to evoke stronger potentials - regardless of the impact this has on the frequency specificity of the test. The work contained in this thesis seeks to rectify this by contributing a new and novel dataset - A series of EEG experiments conducted whilst the participants participated in a Pure-Tone Audiogram-like test, where pure-tones were presented without modulation or modification, as in a standard Pure-Tone Audiogram. This data is then examined as a classification task, to seek to identify whether unmodified pure-tones can evoke a strong enough neural response to be detectable using modern Machine Learning techniques, and examine whether a pure-tone audiogram can be conducted without a participant's response. Finally, novel transforms of this EEG data are examined and statistically analysed to attempt to detect hearing events, including the use of P300 and N1-P2 Peak detection and analysis, plus the extraction and use of Auditory Event Related Synchronisation/Desynchronisation, to examine whether the underlying neural activity can be detected and explained. These results provide a first-of-their-kind examination of Audiometry through the lens of EEG data, allowing us to examine how the brain processes pure-tone auditory stimuli, and opening the door to future research into the use of EEG as a replacement for behavioural audiometry - whilst maintaining the frequency-specificity of a Pure-Tone Audiogram.