Elsevier

NeuroImage

Volume 150, 15 April 2017, Pages 395-404
NeuroImage

Probabilistic conversion of neurosurgical DBS electrode coordinates into MNI space

https://doi.org/10.1016/j.neuroimage.2017.02.004Get rights and content

Highlights

  • Conversion tool between MNI space (used in neuroimaging) and AC/PC coordinates (used in neurosurgical literature).

  • Approach validated using deep brain stimulation electrodes in Parkinson's Disease and Treatment-resistant Depression.

  • Deep brain stimulation target definitions within MNI space across eight diseases.

  • Characterization of deep brain stimulation target for Essential Tremor using multiple subcortical atlases and standardized structural and functional connectomes.

Abstract

In neurosurgical literature, findings such as deep brain stimulation (DBS) electrode positions are conventionally reported in relation to the anterior and posterior commissures of the individual patient (AC/PC coordinates). However, the neuroimaging literature including neuroanatomical atlases, activation patterns, and brain connectivity maps has converged on a different population-based standard (MNI coordinates). Ideally, one could relate these two literatures by directly transforming MRIs from neurosurgical patients into MNI space. However obtaining these patient MRIs can prove difficult or impossible, especially for older studies or those with hundreds of patients. Here, we introduce a methodology for mapping an AC/PC coordinate (such as a DBS electrode position) to MNI space without the need for MRI scans from the patients themselves. We validate our approach using a cohort of DBS patients in which MRIs are available, and test whether several variations on our approach provide added benefit. We then use our approach to convert previously reported DBS electrode coordinates from eight different neurological and psychiatric diseases into MNI space. Finally, we demonstrate the value of such a conversion using the DBS target for essential tremor as an example, relating the site of the active DBS contact to different MNI atlases as well as anatomical and functional connectomes in MNI space.

Introduction

In the field of functional neurosurgery, target locations have been described using coordinates of a defined stereotactic space since 1906 (Clarke and Horsley, 1906). Currently, the Schaltenbrand-Wahren atlas for stereotaxy of the human brain (Schaltenbrand et al., 1977) and the Talairach Co-planar stereotactic Atlas of the Human Brain (Talairach and Tournoux, 1988) serve as standards for reporting brain locations with respect to the anterior commissure (AC) and posterior commissure (PC). Both the AC and PC are small structures that can clearly be identified and are considered relatively invariant in their spatial location (Brett et al., 2002). The largest studies of deep brain stimulation (DBS) for a variety of neurological and psychiatric indications have reported electrode locations in AC/PC coordinates (Table 1). In contrast to neurosurgery, the neuroimaging field has gradually moved away from the single-subject AC/PC standard to population-based atlases. In 1994, the Montreal Neurological Institute (MNI) matched anatomical images of 305 subjects to the Talairach brain (Collins, 1994), which was iteratively refined to the MNI152 2009 NLIN atlas (Fonov et al., 2009). This MNI atlas space has become the standard for reporting results across thousands of neuroimaging studies.

Given these different coordinate system standards, it is difficult to relate findings in the neurosurgical literature (such as clinical DBS response at a given AC/PC coordinate) to findings in the neuroimaging literature (such as activation or connectivity). Relating these two atlas standards is potentially valuable as there are an increasing number of resources available in MNI space that could lend insight into the effect of stimulation at a given brain location (Fox et al., 2014, Horn and Kühn, 2015, Höflich et al., 2010). These MNI resources include subcortical atlases based on histology (Amunts et al., 2013, Chakravarty et al., 2006, Jakab et al., 2012, Krauth et al., 2010, Morel, 2013, Yelnik et al., 2007), high-field MRI (Keuken et al., 2013, Keuken et al., 2014), structural connectivity (Accolla et al., 2014, Behrens, T.E.J., Johansen-Berg, H., Woolrich, M.W., Smith, S.M., Wheeler-Kingshott, C.A.M., Boulby, P.A., Barker, G.J., Sillery, E.L., Sheehan, K., Ciccarelli, O., Thompson, A.J., Brady, J.M., Matthews, P.M., 2003. Non-invasive mapping of connections between human thalamus and cortex using diffusion imaging. 6, 750–757. doi:10.1038/nn1075.) and functional connectivity (Choi et al., 2012, Zhang et al., 2008). Beyond atlases, there are increasingly detailed structural and functional connectome datasets in MNI space (Horn, 2015, Mori et al., 2008, Yeh and Tseng, 2011; Yeo et al., 2011; Setsompop et al., 2013; Van Essen et al., 2012) that can be used to investigate the connectivity properties of DBS targets (Fox et al., 2014) or brain lesions (Boes et al., 2015, Laganiere et al., 2016, Fischer et al., 2016, Darby et al., 2016).

There are several potential options for converting AC/PC coordinates from a neurosurgical study into MNI space. By far the best option is to obtain the MRI data from the neurosurgical patients included in the study and directly warp their brains into MNI space. This allows for direct conversion between each patient's AC/PC coordinates and MNI coordinates. Indeed some neurosurgical studies are beginning to use this approach and report results in MNI space (Barow et al., 2014, Hohlefeld et al., 2015, Horn and Kühn, 2015, Merkl et al., 2016, Merkl et al., 2013, Neumann et al., 2015a, Neumann et al., 2015b, Riva-Posse et al., 2014, Schönecker et al., 2009, Schroll et al., 2015). However, these studies are few relative to the wealth of information in the neurosurgical literature. For example, papers reporting MNI coordinates of DBS sites for Parkinson's disease range from 10–20 patients (e.g. Barow et al., 2014; Neumann et al., 2015), compared to >150 patients for papers reporting AC/PC coordinates (e.g. Caire et al., 2013). Moreover, for most treatment indications, no studies have reported MNI coordinates (Höflich et al., 2010; see Table 1). Obtaining pre and post operative neuroimaging from all these neurosurgical cohorts for direct transformation into MNI space is difficult if not impossible. A conversion tool between AC/PC coordinates and MNI space that does not require the original MRI scans from the neurosurgical patients themselves would be valuable.

However, transforming between coordinate systems is not straightforward. For example, the MNI brain is substantially larger than average (Allen et al., 2002), whereas the Talairach brain is smaller than average (Fig. 1). Talairach-to-MNI conversion tools based on linear (Brett et al., 2002, Lancaster et al., 2007) and nonlinear transforms (Lacadie et al., 2008) were designed to map from Talairach to MNI – not from AC/PC coordinates used in functional surgery. Explicitly, surgical coordinates are often reported relative to the patient's midcommisural point (MCP) or even the PC, requiring an initial conversion into AC-based Talairach-coordinates. This additional conversion step requires knowing the AC-PC distance of the cohort, which is rarely reported (for exceptions see Papavassiliou et al., 2004; Ponce et al., 2015). The AC-PC distance varies between Talairach and MNI space, from 19 to 32 mm across single subjects, and from 24.9 to 28.3 mm across different populations (Fig. 1; Fiandaca et al., 2011; Lee et al., 2008; Liang et al., 2015; Papavassiliou et al., 2004). Moreover, the exact landmarks used to define the AC and PC themselves vary across centers (Weiss et al., 2003; Fig. 1b).

Here, we present a method that converts AC/PC coordinates to MNI space in a probabilistic fashion. In contrast to the solutions mentioned above, mappings are carried out using the individual anatomy in large cohorts of subjects. We validated our approach using two cohorts of DBS patients, one with Parkinson's disease (PD) with DBS to the subthalamic nucleus (STN) and one with Treatment-resistant Depression (TRD) with DBS to the subcallosal cingulate (SCC; Merkl et al., 2015). We chose the PD cohort because the STN is the most common stereotactic target world-wide and spatially close to the AC and PC. We chose the TRD cohort because the subcallosal cingulate is much further from the AC-PC line, helping test for generalizability of our approach. Following validation, we then use our approach to transform average AC/PC coordinates reported in the neurosurgical DBS literature into MNI space. Finally, we demonstrate how using such a conversion allows one to take advantage of MNI-based atlases and tools such as anatomical and functional connectomes to better characterize DBS locations.

Section snippets

Subject cohorts and imaging

450 subjects total from five cohorts were used in this study. The reason for including different cohorts was to determine the relative value of using young healthy subjects, age-matched, disease-matched, or disease severity matched cohorts for our probabilistic mapping.

  • 1.

    Young: 32 young healthy subjects were downloaded from the Human Connectome Project database (mean age 31.5 years±8.6 SD, 14 female, see acknowledgements; Setsompop et al., 2013). 3 subjects of the original 35 were excluded because

Results

To avoid marking the AC and PC locations by hand in all 450 subjects, we first validated an automated method for automatically marking these locations based on non-linear warping of an atlas (Pallavaram et al., 2015). RMS distance errors between manually-marked and automatically-marked coordinates were 0.29 mm (X-Axis), 1.59 mm (Y-Axis) and 1.16 mm (Z-Axis) for the Young cohort (Fig. S1). For DBS patients, results were similar (0.53 mm on X-, 1.27 mm on Y- and 1.33 mm on Z-axis). These errors are on

Discussion

There are three main findings from this study. First, we presented and validated a conversion tool to map from AC/PC coordinates to MNI space in a probabilistic fashion by taking individual anatomical variation into account. Second, we used this tool to identify MNI coordinates for classical DBS targets defined in the literature. Finally, we demonstrate the utility of integrating DBS lead locations with MNI-based resources, using the DBS target for essential tremor as an example. Implications

Conclusions

We introduced a method for converting stereotactic AC/PC coordinates to MNI space in a probabilistic fashion that incorporates anatomic variability. Our method was validated using two cohorts of DBS patients, appears superior to alternative methods, and works well using transforms derived from healthy subject MRIs. We used this method to convert stereotactic coordinates of common DBS targets into MNI space, providing a resource for future studies. Finally, we use the DBS target for essential

Acknowledgements

We would like to thank Thomas Yeo, Dongyang Zhang and Abraham Snyder for helpful advice during the preparation of this manuscript.

AH received funding from Stiftung Charité; Berlin Institute of Health and Prof. Klaus Thiemann Foundation. He received travel stipends from Movement Disorders Society and Ipsen Pharma. AAK was supported by the German Research Agency (DFG - Deutsche Forschungsgemeinschaft). Grant no.: KFO 247 and received honoraria from St Jude Medical and Medtronic; travel grants

References (80)

  • M. Joliot et al.

    AICHA: an atlas of intrinsic connectivity of homotopic areas

    J. Neurosci. Methods

    (2015)
  • M.C. Keuken et al.

    Quantifying inter-individual anatomical variability in the subcortex using 7 T structural MRI

    NeuroImage

    (2014)
  • A. Krauth et al.

    A mean three-dimensional atlas of the human thalamus: generation from multiple histological data

    NeuroImage

    (2010)
  • M. Kuklisova-Murgasova et al.

    A dynamic 4D probabilistic atlas of the developing brain

    NeuroImage

    (2011)
  • C.M. Lacadie et al.

    More accurate Talairach coordinates for neuroimaging using non-linear registration

    NeuroImage

    (2008)
  • A.R. Laird et al.

    Comparison of the disparity between Talairach and MNI coordinates in functional neuroimaging data: validation of the Lancaster transform

    NeuroImage

    (2010)
  • A.M. Lozano et al.

    Probing and regulating dysfunctional circuits using deep brain stimulation

    Neuron

    (2013)
  • A. Merkl et al.

    Antidepressant effects after short-term and chronic stimulation of the subgenual cingulate gyrus in treatment-resistant depression

    Exp. Neurol.

    (2013)
  • S. Mori et al.

    Stereotaxic white matter atlas based on diffusion tensor imaging in an ICBM template

    NeuroImage

    (2008)
  • P. Riva-Posse et al.

    Defining critical white matter pathways mediating successful subcallosal cingulate deep brain stimulation for treatment-resistant depression

    Biol. Psychiatry

    (2014)
  • H. Schroll et al.

    Differential contributions of the globus pallidus and ventral thalamus to stimulus-response learning in humans

    NeuroImage

    (2015)
  • K. Setsompop et al.

    Pushing the limits of in vivo diffusion MRI for the Human Connectome Project

    NeuroImage

    (2013)
  • F.-C. Yeh et al.

    NTU-90: a high angular resolution brain atlas constructed by q-space diffeomorphic reconstruction

    NeuroImage

    (2011)
  • J. Yelnik et al.

    A three-dimensional, histological and deformable atlas of the human basal ganglia. I. Atlas construction based on immunohistochemical and MRI data

    NeuroImage

    (2007)
  • Accolla, E.A., Herrojo Ruiz, M., Horn, A., Schneider, G.-H., Schmitz-Hubsch, T., Draganski, B., Kühn, A.A., 2016. Brain...
  • E.A. Accolla et al.

    Brain tissue properties differentiate between motor and limbic basal ganglia circuits

    Human. Brain Mapp.

    (2014)
  • L. Ackermans et al.

    Double-blind clinical trial of thalamic stimulation in patients with Tourette syndrome

    Brain

    (2011)
  • J.S. Allen et al.

    Normal neuroanatomical variation in the human brain: an MRI‐volumetric study

    Am. J. Phys. Anthropol.

    (2002)
  • K. Amunts et al.

    BigBrain: an ultrahigh-resolution 3D human brain model

    Science

    (2013)
  • D. Anderson et al.

    Treatment of patients with intractable obsessive—compulsive disorder with anterior capsular stimulation

    J. Neurosurg.

    (2003)
  • J.S. Anderson et al.

    Functional connectivity targeting for deep brain stimulation in essential tremor

    AJNR Am. J. Neuroradiol.

    (2011)
  • Ashburner, J., 2007. A Fast Diffeomorphic Image Registration Algorithm. vol. 38, pp. 95–113....
  • E. Barow et al.

    Deep brain stimulation suppresses pallidal low frequency activity in patients with phasic dystonic movements

    Brain

    (2014)
  • Behrens, T.E.J., Johansen-Berg, H., Woolrich, M.W., Smith, S.M., Wheeler-Kingshott, C.A.M., Boulby, P.A., Barker, G.J.,...
  • A.D. Boes et al.

    Network localization of neurological symptoms from focal brain lesions

    Brain: J. Neurol.

    (2015)
  • M. Brett et al.

    The problem of functional localization in the human brain

    Nat. Rev. Neurosci.

    (2002)
  • F. Caire et al.

    A systematic review of studies on anatomical position of electrode contacts used for chronic subthalamic stimulation in Parkinson's disease

    Acta Neurochir.

    (2013)
  • V. Chan-Palay

    Cerebellar Dentate Nucleus: Organization, Cytology and Transmitters

    (1977)
  • E.Y. Choi et al.

    The organization of the human striatum estimated by intrinsic functional connectivity

    J. Neurophysiol.

    (2012)
  • R.H. Clarke et al.

    On a method of investigating the deep ganglia and tracts of the central nervous system (cerebellum)

    Br. Med. J.

    (1906)
  • Cited by (108)

    • Interrater reliability of deep brain stimulation electrode localizations

      2022, NeuroImage
      Citation Excerpt :

      To investigate this correlation and further explore location-specific DBS-effects in scientific studies, it is crucial to exactly reconstruct electrode placement. The open-source toolbox Lead-DBS was developed with this aim in 2012 by the last author of the present study (Horn and Kühn, 2015a) and has since continuously been optimized by investigators from various institutions (Dembek et al., 2019; Ewert et al., 2018, 2019; Horn et al., 2017; Horn et al., 2019a; Horn and Kühn, 2015b; Husch et al., 2018; Treu et al., 2020). In brief, Lead-DBS facilitates semiautomatic localization based on structural neuroimaging.

    View all citing articles on Scopus
    View full text