Cart
0
Vagus nerve stimulation (VNS) has been used in the treatment of epilepsy and depression for more than 20 years. Although the invasive cervical method is the most preferred application, side effects such as cough, voice change and hoarseness can be seen due to negative effects on the recurrent laryngeal nerve (a branch of the vagus nerve). Auricular VNS has been preferred recently due to its non-invasiveness, but uncertainty about the stimulation parameters continues. We tested the hypothesis that auricular VNS can affect voice and its features indirectly via afferent nerve connections reaching the nucleus tractus solitarius. Two patients previously using auricular VNS device for different diseases were requested to record their voices before and after the stimulation. Their devices (Vagustim) were changed with new version to check the usage of the patients. Sound recordings at different VNS frequencies (1-150 Hz) were collected by a mobile phone and analyzed with Praat and our MATLAB algorithm. Fundamental frequency (f0), jitter, shimmer, and harmonic to noise ratio (HNR) owere evaluated. The alteration was highest at 100 Hz and 30 Hz VNS for the male and female patients respectively. Audio recordings before and after 30 Hz (for female) and 100 Hz (for male) VNS at different durations (5-30 min) on different days were repeated and compared by Praat and our algorithm. Some discrepancy between the parameters jitter, shimmer, and HNR are detected between the algorithms, which is accounted to the fact that it is not standardized whether the algorithm uses only a specific part of the input signal or the whole signal. However, when the ratio of change of these parameters are considered, fundamental frequency and the HNR were found to be highly consistent for developing an algorithm to govern the stimulation parameters in an automated way. Furthermore, the same ratios for jitter and shimmer are also promising after some improvement to be included in such an algorithm. These results suggest that auricular VNS can affect voice and its parameters, but this change is related with stimulation parameters. It seems necessary to develop specific software and algorithms that can detect this change well.
Authors: Ali Veysel Özden, Tayfun Tatar
