Elsevier

Clinical Neurophysiology

Volume 124, Issue 1, January 2013, Pages 107-113
Clinical Neurophysiology

Feasibility of clinical Magnetoencephalography (MEG) functional mapping in the presence of dental artefacts

https://doi.org/10.1016/j.clinph.2012.06.013Get rights and content

Abstract

Objective

To evaluate the viability of MEG source reconstruction in the presence of large interference due to orthodontic material.

Methods

We recorded the magnetic fields following a simple hand movement and following electrical stimulation of the median nerve (somatosensory evoked field –SEF). These two tasks were performed twice, once with and once without artificial dental artefacts. Temporal Signal Space Separation (tSSS) was applied to spatially filter the data and source reconstruction was performed according to standard procedures for pre-surgical mapping of eloquent cortex, applying dipole fitting to the SEF data and beamforming to the hand movement data.

Results

Comparing the data with braces to the data without braces, the observed distances between the activations following hand movement in the two conditions were on average 6.4 and 4.5 mm for the left and right hand, respectively, whereas the dipole localisation errors for the SEF were 4.1 and 5.4 mm, respectively. Without tSSS it was generally not possible to obtain reliable dipole fit or beamforming results when wearing braces.

Conclusion

We confirm that tSSS is a required and effective pre-processing step for data recorded with the Elekta-MEG system. Moreover, we have shown that even the presence of large interference from orthodontic material does not significantly alter the results from dipole localisation or beamformer analysis, provided the data are spatially filtered by tSSS.

Significance

State-of-the-art signal processing techniques enable the use of MEG for pre-surgical evaluation in a much larger clinical population than previously thought possible.

Highlights

► State-of-the-art MEG signal processing techniques enable the removal of artefacts introduced by orthodontic braces. ► Localisation accuracy of SEF and motor responses is comparable in subjects wearing and not wearing braces. ► MEG can be used for pre-surgical evaluation in a much larger clinical population than previously thought possible.

Introduction

Whole-head Magnetoencephalography (MEG) directly and non-invasively measures electromagnetic activity from neurons in the brain. Although MEG requires specialized instrumentation and procedures, it can be used routinely in a clinical setting to record responses from children and adults (Stufflebeam et al., 2009). In particular, MEG has become an important tool for pre-surgical mapping of eloquent cortex in surgical candidates with brain tumours and/or epilepsy (Seo et al., 2011, Pirmoradi et al., 2010), involving typically voluntary movements and somatosensory stimulation. The latest advances in source modelling procedures, such as beamforming, enable the localisation of healthy brain tissue involved in, for example, voluntary movement with exquisite temporal (∼1 ms) and good spatial resolution (1–20 mm) (Hillebrand and Barnes, 2005). However, a common problem with MEG is the presence of interference due to external sources (e.g. power line interference, cars) and from the participant itself (e.g. heart and muscle interference, orthodontic material) (Vrba, 2002). Modern MEG systems aim to reduce the magnitude of such interferences using a combination of hardware filtering (axial or planar gradiometer configurations) and spatial filtering in software (by forming higher-order gradiometers (Vrba et al., 1999) or signal-only subspaces (Taulu et al., 2004)).

Additional noise rejection can be achieved through signal averaging, or through beamformer based source reconstruction (Cheyne et al., 2007, Adjamian et al., 2009). In a clinical setting, artefacts caused by the presence of metallic orthodontic material, such as braces, can be of such magnitude that it becomes impossible to record an interpretable MEG in some patient groups, like young people, who relatively often wear dental braces nowadays: in our clinical MEG population the incidence of orthodontic appliances was approximately 8% over the last 2 years, which is in general agreement with epidemiological data from the UK (Chestnutt et al., 2006). Fortunately, with recent developments in signal processing, namely the Signal Space Separation (SSS) method with temporal extension (tSSS) (Taulu and Simola, 2006), it is possible to remove even these large artefacts, albeit at the cost of potentially removing true brain signals as well (Medvedovsky et al., 2009). It is therefore important to assess the effect of tSSS on the accuracy of subsequent source reconstructions, particularly when large interference is present. In this study we aim to show that tSSS in combination with beamforming or dipole fitting (using averaged signals) leads to accurate source reconstructions, even in the presence of large interference from orthodontic material. We recorded MEG data during simple voluntary hand movements, as well as following electrical stimulation of the median nerve, both with and without dental artefacts, and used tSSS to spatially filter the data. We subsequently localised the brain regions that were activated during the movement task using beamforming and applied dipole fitting to the SEF data. To the best of our knowledge this is the first systematic study on the effects of dental artefacts on the accuracy of reconstructed neuronal activity and the use of tSSS to improve the localisation accuracy.

Section snippets

Participants

Six healthy right handed volunteers (4 male, 2 female; mean age: 45 years; age range 33–54) were recruited from staff at the VU University Medical Center (Amsterdam, The Netherlands). Five of the participants did not have dental work that caused discernable artefacts in the MEG. One participant used a brace in daily life. All the participants provided informed consent, and the study was approved by the Medical Ethics Committee of the VU University Medical Center.

Metallic dental artefacts

Orthodontic braces for the upper

Signal Space Separation

To remove artefacts, the raw data were spatially filtered offline using the temporal extension of Signal Space Separation (tSSS) (Taulu and Simola, 2006, Taulu and Hari, 2009), using MaxFilter software (Elekta Neuromag Oy, version 2.1). SSS decomposes the data into a set of spherical harmonics and divides these basis functions in those coming from the head (brain signals) and those not coming from the head (noise signals). The coefficients for these basis functions are estimated for sections of

Results

For all subjects we observed large artefacts in the ongoing MEG when braces were worn. Typically, these artefacts were modulated by breathing and cardiac rhythms. The subjects that wore temporary braces showed more artefacts due to swallowing or tongue movements than the subject with ordinary fixed braces. Movement of the muscles of mastication causes myogenic activity. Similarly, excessive salivation induces deglutition and related tongue movement, also causing myogenic activity (Furlong et

Discussion

In this study we have shown that even the presence of large interference from orthodontic material does not significantly alter the results from dipole localisation (after signal averaging) or beamformer analysis, provided the data are spatially filtered by tSSS. The localisation errors introduced by the interference from braces were in the order of 0.5 cm for oscillatory responses following hand movement as well as for the sources underlying SEFs (as compared to the condition when no braces

Acknowledgments

The authors would like to thank PeterJan Ris, Ndedi Sijsma and Karin Plugge, who performed the MEG recordings, as well as Prof. Herman Van Beek (Academic Centre for Dentistry Amsterdam) for his support with construction of the orthodontic braces. Author P.F. was supported by a grant from AICE (Italian Association Against Epilepsy).

References (33)

  • J. Vrba

    Magnetoencephalography: the art of finding a needle in a haystack

    Physica C

    (2002)
  • J. Vrba et al.

    Signal processing in magnetoencephalography

    Methods

    (2001)
  • I.G. Chestnutt et al.

    The orthodontic condition of children in the United Kingdom 2003

    Br Dent J

    (2006)
  • W.C. Gaetz et al.

    Presurgical localisation of primary motor cortex in pediatric patients with brain lesions by the use of spatially filtered magnetoencephalography

    Neurosurgery

    (2009)
  • A. Hillebrand et al.

    A new approach to neuroimaging with magnetoencephalography

    Hum Brain Mapp

    (2005)
  • A. Korvenoja et al.

    Sensorimotor cortex localization: comparison of magneto encephalography, functional MR imaging, and intraoperative cortical mapping

    Radiology

    (2006)
  • Cited by (46)

    • Comparing the potential of MEG and EEG to uncover brain tracking of speech temporal envelope

      2019, NeuroImage
      Citation Excerpt :

      MEG recordings may prove challenging in participants wearing cochlear implant, teeth braces or any metallic devices, because all these generate strong interferences. Mitigating this limitation, most of these nearby interferences can be efficiently subtracted from the MEG signals of interest with appropriate preprocessing (Bourguignon et al., 2016; Carrette et al., 2011a; Hillebrand et al., 2013; Mäkelä et al., 2007; Medvedovsky et al., 2009; Taulu and Hari, 2009; Taulu and Simola, 2006). As another limitation, MEG helmets do not adapt to participants’ head size, which implies that individuals with exceptionally large head do not fit in the helmet, and more importantly, when assessing children, the large sensor–brain distance may potentially diminish the SNR (Wehner et al., 2008).

    • Interactions between frontal and posterior oscillatory dynamics support adjustment of stimulus processing during reinforcement learning

      2018, NeuroImage
      Citation Excerpt :

      An anterior source-level cluster also demonstrated larger theta power after negative feedback, in line with EEG research showing that aPFC theta power reflects feedback valence and predicts behavioral adaptation and exploration (Cavanagh et al., 2012a,b; van de Vijver et al., 2014). Note, however, that eye blinks were not removed for the source-level analyses, although the tSSS and beamforming procedures likely removed most of this artifact (Adjamian et al., 2009; Hillebrand et al., 2013; but see Hipp and Siegel, 2015). Indeed, control analyses confirmed that eye blink artifacts did not affect our results.

    View all citing articles on Scopus
    View full text