Electrophysiological characterization of tinnitus in listeners with normal audiogram and supra-threshold hearing deficits

Tinnitus is the perception of sound when no external source is present. Despite the high prevalence of this phenomenon (estimated to be 1/10 people in the USA), there are still many open questions. For example, there is no standard definition to discern between temporary ringing in the ear after a triggering event and the regularly recurring one. Moreover, no treatment available is efficient for all the tinnitus sufferers. The objective of this thesis is to evaluate new and old tools for detecting and uncovering new aspects of tinnitus. The evaluation began with machine learning, aiming to categorize tinnitus subgroups by collecting a large number of diverse measures for a tinnitus and a control group, and using a clustering algorithm. It was hypothesized one cluster would have characteristics close to the ones hypothesized for cochlear synaptopathy (CS). CS is the disruption of synapses between the inner hair cells and the auditory nerve and is hypothesized to be a possible cause for tinnitus in participants with no hearing loss. The measures chosen belonged to one of the following areas: psychoacoustic (Acoustic Categorical Loudness Scaling, ACALOS; Psychophysical Tuning Curve, PTC; Tinnitus Tuning Curve, TTC; High Frequency audiometry, HFA; tinnitus likeness, TL), physiology (Middle Ear Muscle Reflex, MEMR; MEMR response latency), questionnaires (Tinnitus Handicap Inventory, THI), and general information (age). Interestingly, a cluster with the characteristics theorized for CS was detected. However, the number of clusters was high with respect to the number of participants (5 for 35) making it difficult to draw conclusions. In agreement to the literature, it had been hypothesized there would be a higher relevance for MEMR and HFA, but this was not the case. The results for MEMR and HFA then motivated our following studies. We continued with testing the MEMR with a focus on hyperacusis, hypothesizing a possible confound between the two. The study was conducted with not only the classical artificial sounds, but also natural sounds. However, the
study did not show a correlation between hyperacusis and the MEMR strength. Interestingly, the measured MEMR strength was different for different sounds, as well as the perceived unpleasantness. This find could be of interest for testing participants particularly sensitive to sounds. We then proceeded to investigate the HFA method used in the categorization study. The method selected in the data collection was not the classical one, but optimized with a Bayesian algorithm. The algorithm was chosen to speed up the collection of the data. However, we found this method to be reliable and with errors close to the expected ones. With no measures in the initial study directly analyzing the central system, for the final study we elected to see how the brain waves change with and without tinnitus. To experiment that, we selected the delta and alpha frequency
bands, whose powers were found to be enhanced and reduced respectively in previous studies. In particular, we used the Dynamical Bayesian Inference method to extract the coupling function and the coupling strength between the two frequency bands. The coupling function gives information about the interaction between the two oscillators, while the coupling strength gives information about the prevalent direction of the coupling. The data collected included the resting state EEG for a tinnitus and a control group and the resting state MEG for two conditions: control and induced tinnitus, in the same subjects. As observed in previous literature, the EEG showed that the stronger coupling was from delta to alpha, but when comparing the tinnitus and control group there were significant differences in the coupling strength for both the directions. The coupling functions showed also different behaviors between the two groups. For the MEG there were significant differences between the two conditions only in A1 from alpha to delta.