Tinnitus-related Abnormalities in Brain Networks During Stimulated and Resting States

Cris Lanting, Aron Woźniak, Pim van Dijk, Dave Langers

We investigated tinnitus-related differences in functional networks in adults with tinnitus in a functional connectivity study. Previously it was found that various networks show differences in connectivity in patients with tinnitus compared to controls. Whether this relates to patients’ ongoing tinnitus and whether the ecological sensory environment modulates connectivity remains unknown

Twenty healthy controls and twenty patients suffering from chronic tinnitus were enrolled in the study. Except for the presence of tinnitus in the patient group, all subjects were selected to have normal or near-normal hearing. fMRI data were obtained in two different functional states. In one session subjects freely viewed emotionally salient movie fragments (“fixed-state”) while in the other they were not performing any task (“resting-state”).

After data pre-processing, Principal Component Analysis was performed to obtain 25 components for all datasets. These were fed into an Independent Component Analysis (ICA), concatenating the data across both groups and both datasets, to obtain eleven group-level networks of neural origin, each consisting of spatial maps with their respective time-courses. Subject-specific maps and their time-course were obtained by back-projection (Dual Regression). For each of the components a mixed-effects linear model was composed with factors group (tinnitus vs. controls), task (fixed-state vs. resting state) and their interaction.

The neural components comprised the visual, sensorimotor, auditory, and limbic systems, the default mode, dorsal attention, executive-control, and frontoparietal networks, and the cerebellum. Most notably, the default mode network (DMN) was less extensive and significantly less strongly active in tinnitus patients than in controls. This group difference existed in both paradigms. At the same time, the DMN was stronger during resting-state than during fixed-state in the controls but not the patients. We attribute this pattern to the unremitting engaging effect of the tinnitus percept.