Acta Neurochirurgica

, Volume 154, Issue 3, pp 509–515

Assessment of non-motor hearing symptoms in hemifacial spasm using magnetoencephalography


  • Young Seok Park
    • Department of NeurosurgeryYonsei University College of Medicine
    • Department of Neurosurgery, Bundang CHA Medical CenterCHA University
  • Bong Soo Kim
    • Department of NeurosurgeryYonsei University College of Medicine
  • Dong Kyu Lee
    • Department of NeurosurgeryYonsei University College of Medicine
  • Seung-Koo Lee
    • Department of Radiology, Severance Hospital, MEG Center, Severance Hospital Brain Korea 21 Project for Medical Science, Brain Research InstituteYonsei University College of Medicine
  • Hyuk Chan Kwon
    • Center for Brain and Cognitive Science ResearchKorea Research Institute of Standards and Science
  • Kiwoong Kim
    • Center for Brain and Cognitive Science ResearchKorea Research Institute of Standards and Science
  • Yong Ho Lee
    • Center for Brain and Cognitive Science ResearchKorea Research Institute of Standards and Science
    • Department of NeurosurgeryYonsei University College of Medicine
Clinical Article

DOI: 10.1007/s00701-011-1231-y

Cite this article as:
Park, Y.S., Kim, B.S., Lee, D.K. et al. Acta Neurochir (2012) 154: 509. doi:10.1007/s00701-011-1231-y



Hemifacial spasm patients often suffer from non-motor symptoms such as tinnitus. These non-motor symptoms are known to be associated with changes in cortical activity. Magnetoencephalography (MEG) is a technique that can record brain activity noninvasively. To determine the usefulness of MEG in assessing changes in cortical activity associated with non-motor hearing symptoms in hemifacial spasm patients.


We used MEG to evaluate the reactivity of the auditory cortex in 26 hemifacial spasm patients. We divided patients into a subjective tinnitus group (n = 10) and a non-tinnitus group (n = 16). The latency and amplitude of the most prominent deflection, N100m, was compared between the two groups.


There was a significant difference in the pure tone audiogram on the spasm side compared with the non-spasm side. After stimulation on the spasm side, the amplitude of the N100m peak in the contralateral hemisphere was lower in the subjective tinnitus group than in the non-tinnitus group.


Our results indicate that MEG can detect differences in cortical activity between hemifacial spasm patients with and without tinnitus. This suggests that MEG can identify changes in cortical activity associated with non-motor symptoms.


DysynchronizationHearingHemifacial spasmMEG


Non-motor symptoms such as tinnitus [34], acoustic abnormalities and loss of hearing [18] are frequently found in hemifacial spasm (HFS) patients. The close anatomical relationship between the facial and auditory nerves makes it likely that a vascular abnormality compressing the facial nerve will also affect the auditory nerve. This could lead to measurable changes in auditory function [18]. Magnetoencephalography (MEG) can record spontaneous auditory activity noninvasively. Auditory evoked magnetic fields (AEF) are derived from spontaneous cortical sources in the lower auditory levels, and MEG can detect changes in the lower auditory levels in the supratemporal auditory cortex. The most prominent deflection, N100m, peaks 100 ms after the onset of the sound. The N100m is typically larger and has a shorter latency (4–14 ms shorter) when the contralateral (versus ipsilateral) ear is stimulated. The amplitude of N100m, in both the ipsilateral and contralateral hemisphere, increases as the interstimulus interval (ISI) of the sound increases. The amplitude and latency of N100m are highly reproducible [31].

Approximately 5–15% of the population have an auditory phantom sensation, termed tinnitus, in the absence of external sound [3]. Tinnitus associated with HFS is not uncommon. Tinnitus associated with HFS is very likely due to neurovascular compression and will disappear after neurovascular decompression [29]. It was not caused by irritation of the stapedius nerve, but the eighth cranial nerve was simultaneously compressed by offending vessels. Whether it is caused by some irritation of the eighth nerve or multilple level of projection of hearing pathway, results in cortical reorganization with spontaneous neural activity changes.

To assess the usefulness of MEG in detecting the non-motor hearing symptom of tinnitus in HFS patients, we investigated the the latency and amplitude of N100m in these patients.

Materials and methods


From November 2009 to August 2010, 26 patients (16 women, 10 men) with HFS were examined via MEG at Severance Hospital, Seoul, Korea. All MEGs occurred prior to any surgical treatment. The median patient age was 49.0 years (range, 25–69 years). Patients were divided into two groups: a subjective tinnitus group (n = 10) and a non-tinnitus (n = 16) group (Table 1).
Table 1

Clinical characteristics of the study subjects



Age: median (range) in years

49 (25–69)

Sex: female/male


Disease duration: median (range) in years

5 (1–10)

Side: right/left


Offender: AICA/PICA


Tinnitus: yes/no


AICA anterior inferior cerebellar artery, PICA posterior inferior cerebellar artery

All subjects underwent audiological inspection, pure tone audiometry, and magnetic resonance imaging (MRI). To be included in this study, patients had to have unilateral involuntary facial muscle contractions that affected one or more muscle groups innervated by the facial nerve. Patients with myokymia (focal, undulating muscle contraction), tardive dyskinesia, or other forms of facial or oromandibular dystonic movements were excluded. We included patients only if an offending vessel was observed via preoperative three-dimensional time-of flight magnetic resonance angiography imaging. Three-dimensional MRI T1-weighted images and T1–weighted gadolinium enhanced images were employed to ensure the absence of brain structure abnormalities in the study participants [parameters used: TR (repetition time) 13.6 ms, TE (echo time) 4.8 ms, recording matrix 256 × 256 pixels, one excitation, 240-mm field of view, 2-mm slice thickness]. Only patients who had not had previous surgery to relieve HFS were included in this study. All participants provided written informed consent prior to enrollment in the study. The institutional review board of Severance Hospital approved this study (6-2009-0156).


Ten of the enrolled patients had tinnitus, a non-motor hearing symptom often associated with HFS. Thorough clinical evaluation by an otologist or otolaryngologist allowed us to exclude objective tinnitus and/or other ear disease as the cause of the tinnitus in these patients. We did not perform tinnitus pitch match, tinnitus loudness, or digital tinnitus tests. Therefore, no assessment of the severity of tinnitus was made.


Tone bursts of 100-ms duration (15-ms slope) were used for acoustic stimulation. Pure tones of 1 kHz were applied to either the left and right ear. Each block of frequencies consisted of 100 epochs of an interstimulus interval ((ISI) of 900–1,000 ms. Tones were delivered at a comfortable sound pressure level (90 dB above threshold) through two 2.5-m-long silicon tubes connected to earphones (ER-30, Etymotic Research, Elk Grove Village, IL, USA). Auditory stimuli were generated by the STIM2 system (Compumedics Neuroscan, El Paso, TX, USA).

MEG analysis

MEG recordings of all subjects were obtained using a MEG system developed by the Korean Research Institute of Standards and Science (KRISS, Daejeon, Korea). This MEG system is located in the neurosurgical center of Severance Hospital and has 152 axial first-order double relaxation oscillation (DROS) superconducting quantum interference device (SQUID) gradiometer sensors on a helmet-shaped surface that covers the entire entire scalp [12]. In this study, the 152-channel MEG system was used in a magnetically shielded room. The average noise spectral density in the magnetically shielded room was 10 fT/√Hz at 1 Hz and 5 fT/√Hz at 100 Hz.

Auditory stimulus-triggered epochs of 500-ms duration (including a 100-ms prestimulus baseline) were filtered online with a bandpass of 0.1-100 Hz and recorded at a sampling rate of 500 Hz. The MEG responses to auditory stimuli were averaged to improve the signal-to-noise ratio. The averaged waveforms were filtered offline with a low-pass at 40 Hz, and the baseline for the waveforms was defined as the mean amplitude between −100 and 0 ms relative to tone onset. The peak latency and amplitude of the N100m component were determined for each hemisphere at the time point at which the absolute value of the predefined left/right channels reached a maximum between 80 and 140 ms after the auditory stimulus onset. For isofield mapping, spatial analysis of the averaged waveform, expressed as the root mean square (RMS) over all channels, was conducted after stimulation of both the spasm side and the non-spasm side.

Statistical analysis

Statistical significance in the delay of waves was determined by analysis of variance, and the one-sample t-test was used to compare latency and amplitude between groups. Statistical significance was set to P < 0.05. All statistical analysis was performed via the statistical software program SPSS (Chicago, IL, USA).


Latencies and N100m peak amplitude

A significant difference was found between the pure tone audiometry (PTA) on the spasm side (15.6 ± 12.7 dB) and the PTA on the non-spasm side (12.1 ± 6.0 dB). The detected latency of the N100m peak in the contralateral hemisphere was 112.8 ± 18.9 ms after stimulation of the spasm side and 109.2 ± 9.0 ms after stimulation of the non-spasm side. There was a significant difference between spasm side stimulation versus non-spasm side stimulation in the difference between the latencies of the N100m peaks of the two hemispheres (ipsilateral and contralateral). Figure 1 shows the amplitude and latency of N100m after healthy side and spasm side stimulation.
Fig. 1

A 52-year-old female patient with left hemifacial spasm and subjective tinnitus. The amplitude and latency of N100m after healthy side (a) and spasm side (b) stimulation

The detected amplitude of the N100m peak in the contralateral hemisphere after stimulation of the spasm side was 154.3 ± 44.8 fT, whereas it was 145.1 ± 54.9 fT after stimulation of the non-spasm side. There was a significant difference in the amplitude of N100m in both the ipsilateral and contralateral hemispheres between spasm side stimulation and non-spasm side stimulation. Table 2 shows the mean (± standard deviation) latencies and the source strength of ECDs of the N100m for HFS patients.
Table 2

Comparison of the N100m latency and amplitude of the healthy side versus the spasm side after stimulation of either side


Contralateral hemisphere

Ispilateral hemisphere

P value



15.6 ± 12.7

12.1 ± 6.0


Spasm-side Stimulation

Latency (ms)

112.8 ± 18.9

102.2 ±16.0



Amplitude (nAm)

154.3 ± 44.8

178.5 ± 64.7


Healthy-side stimulation

Latency (ms)

109.2 ± 9.0

99.4 ± 9.8



Amplitude (nAm)

145.1 ± 54.9

185.8 ± 43.0


RMS data revealed a typical N100m band for the non-spasm side, while a broad low amplitude band was observed for the spasm side. The RMS data from the auditory response had a desynchronized pattern on the spasm side (Fig. 2).
Fig. 2

Root mean square (RMS) data showing superimposed waveforms of the data measured on the healthy (a) and spasm sides (b). A broad desynchronization pattern was detected on the spasm side (b)

Comparison of tinnitus and non-tinnitus groups

Ten patients had subjective tinnitus on the spasm side, whereas 16 patients did not. There were no significant differences in PTA between the tinnitus and non-tinnitus groups in either the spasm side (tinnitus group, 19.0 ± 19.5 dB; non-tinnitus group, 13.5 ± 5.4 dB) or the non-spasm side (tinnitus group, 10.3 ± 5.9 dB; non-tinnitus group, 13.2 ± 6.0 dB).

The detected amplitude of the N100m peak in the contralateral hemisphere after spasm side stimulation was significantly different between the two groups: 139.1 ± 36.7 fT in the tinnitus group versus 203.1 ± 67.0 fT in the non-tinnitus group. The detected amplitude of the N100m peak at the contralateral hemisphere after stimulation of the spasm side was significantly different between the two groups: 139.1 ± 36.7 fT in the tinnitus group versus 203.1 ± 67.0 fT in the non-tinnitus group (Table 3).
Table 3

Comparison of the latency and amplitude of the N100m peak between tinnitus and non-tinnitus groups






P value




(n = 10)

(n = 16)




Spasm side

19.0 ± 19.5

13.5 ± 5.4



Healthy side

10.3 ± 5.9

13.2 ± 6.0


Spasm side


Latency (ms)

118.2 ± 28.2

109.4 ± 9.4



Amplitude (nAm)

142.6 ± 34.4

161.6 ± 49.8




Latency (ms)

108.1 ± 23.5

98.6 ± 7.7



Amplitude (nAm)

139.1 ± 36.7

203.1 ± 67.0


Healthy side


Latency (ms)

111.8 ± 8.8

107.6 ± 9.1



Amplitude (nAm)

133.6 ± 30.7

152.4 ± 65.7




Latency (ms)

98.8 ± 8.3

100.2 ± 12.3



Amplitude (nAm)

169.6 ± 54.1

195.9 ± 32.3



In this study, we have demonstrated the feasibility of MEG as a means to detect changes in cortical activity associated with non-motor auditory symptoms in HFS patients. Furthermore, our study suggests that following a peripheral lesion (as occurs in HFS), integration of neuronal activity in the auditory cortex correlate with non-motor symptoms such as tinnitus.

Patients with HFS frequently complain of hearing associated symptoms such as unilateral or bilateral hearing loss (13%), a transient loss of hearing during the active spasm of the facial muscles on the ipsilateral side (0.4%), and a “clicking” or a “ticking” sound in the ipsilateral ear (4%) [32]. However, these complaints do not always correlate with the side or severity of the HFS. The offending vessel in HFS irritates the cochlear and/or vestibular nerves causing a changing in the firing pattern of neurons in the auditory cortex. In HFS, this can lead to a change in spontaneous oscillatory activity without external stimuli. Our data support the hypothesis that synchronous neuronal activity of cells within the auditory cortex could be responsible for tinnitus [33].

Several studies have suggested that the site where the abnormal activity is generated is centered in the area of vascular compression, likely at the facial nucleus [11, 15, 17]. It has been proposed that aberrant afferent activity at the site of nerve compression causes reorganization of the facial nucleus by a mechanism similar to the “kindling” phenomenon [16]. The correlation between the severity of HFS and the number of HFS-accompanying symptoms supports the hypothesis that hyperactivity of the seventh nerve in HFS patients involves not only motor fibers but also sensory and autonomic fibers [27]. The abnormal activity of sensory and autonomic fibers may be generated locally by “ectopic” or “ephaptic” transmission at the compression site [6, 20] or centrally by the spreading of impulses via intraneuronal connections in the brain stem from a hyper-excitable facial nucleus to the parasympathetic nucleus salivarius superior and the sensory nucleus of the solitary tract. Vascular compression of the vestibular cochlear nerve may lead to different symptoms, including tinnitus, hearing loss, disabling vertigo, and imbalance [35]. Tinnitus and vertigo can be alleviated by microvascular decompression (MVD) of the eighth cranial nerve [36]. Neurovascular compression causes dysmyelination and demyelination of the underlying cochlear nerve, resulting in abnormal signal transmission to the auditory cortex [7, 28]. This alteration in signal transmission could lead to reorganization of the auditory tract and auditory cortex, and this reorganization can cause tinnitus [8, 9, 23, 26]. Tinnitus is accompanied by a change of the tonotopic map in the auditory cortex [19, 25]. Tinnitus can be treated with electrical stimulation. The success of this treatment varies, but generally, the more central the stimulation, the better the results [1, 2]. The mechanism of this treatment is thought to be that stimulation indirectly reflects the cortical reorganization in HFS patients with tinnitus. Much like there is a very strong correlation between the amount of reorganization of the somatosensory cortex and the amount of phantom limb pain [4], there is a strong positive association between the magnitude of the subjective tinnitus and the magnitude of the shift in the auditory cortex [19].

Tinnitus that is associated with ipsilateral HFS and (not due to irritation of the stapedius nerve) is very likely to disappear after microvascular decompression [29, 30]. There has been a paucity of evidence to support the contention that tinnitus originates from specific locations along the neural pathway from the cochlear nerve to the brain stem. Using MEG, Llinas et al. [13] showed that patients’ unilateral tinnitus perceptions were accompanied by low- and high-frequency electromagnetic activity that was spontaneous and abnormal and localized to the contralateral auditory cortex. Neuromagnetic auditory responses (waveform, isofield mapping) on the HFS side showed a different pattern from that observed on the healthy side. For example, N100m had a broad band on the spasm side. Our MEG results suggest that it is feasible that peripheral irritation influences the cortex through a neural pathway involving the brain stem.

Muhlnickel et al. [19] found that the cortical source of the N100m magnetic field evoked by tinnitus was shifted on average 2.7 mm from its location in the cortical place map of normal hearing controls. Disturbances of tonotopy that are associated with tinnitus may reflect the neural dynamics that underlie this condition. Map reorganization may reflect an unmasking of either lateral connections in the affected frequency region of the cortical place map [3] or normally silent thalamocortical projections from nearby unaffected frequency regions of the auditory thalamus [21].

Repetitive transcranial magnetic stimulation (rTMS) for tinnitus is an emerging, noninvasive technique that allows cortical activity to be modified as activity can be enhanced or inhibited both at the site of stimulation site as well as at sites downstream (providing there is connectivity in the neural pathway) [5, 10, 14, 22]. The effects of cortical stimulation on ongoing brain activity were investigated using a MEG sensor, and it was found that cortical stimulation significantly reduced ongoing brain activity at lower frequencies (<40 Hz) and increased it at higher frequencies (>40 Hz), indicating the presence of gamma range components [24]. Using this reverse “cortical reorganization” by external stimulation, MEG can provide information helpful in planning a treatment of rTMS or the surgical implantation of a cortical stimulation device in tinnitus patients.

This study has some limitations. First, spontaneous, fine, fascicular muscle contractions could not be excluded. Second, tinnitus, hearing loss, and age tend to be confounded. Hence, it is conceivable that the frequency gradients seen in tinnitus subjects are not related to the presence of tinnitus, but instead to one or both of the aforementioned associated changes. Third, we did not perform the tinnitus pitch match or tinnitus loudness match tests. Therefore, the severity of tinnitus was not assessed in our study. Fourth, we did not have a sham stimulation group in our study. Hence, any placebo effects of the stimulation would not be adequately controlled for. In the future, the parameters of MEG for HFS should be quantified and spatial analysis should be conducted to detect dipole changes. Also, neuromagnetic auditory response analysis would be useful for further investigating the non-motor symptoms of HFS.


In patients with HFS and tinnitus, after spasm side-stimulation, the amplitude of the N100m peak in the contralateral hemisphere was lower than the peak observed in HFS patients without tinnitus. Neuromagnetic auditory responses (waveforms, isofield maps) showed a pattern of dysynchronization on the spasm side. These results clearly demonstrate that MEG can detect cortical activity associated with non-motor symptoms in HFS patients. Furthermore, our MEG results suggest that peripheral irritation influences the cortex through neural pathways involving the brain stem. In summary, this MEG study sheds some light on potential mechanisms of and treatment modalities for the non-motor symptoms of HFS.


We would like to thank Eun Jeong Kwon, RN, Sang Keum Park, RN, and Ester PohelKing, MD, for their assistance in preparing manuscript. This study was supported by a faculty research grant from Yonsei University College of Medicine (6-2009-0156).

Conflicts of interest


Copyright information

© Springer-Verlag 2011