Technical NoteA new method based on ICBM152 head surface for probe placement in multichannel fNIRS
Introduction
Functional near infrared spectroscopy (fNIRS) is a non-invasive neuroimaging technique used since the early 1990s (e.g., Villringer et al., 1993) to investigate hemodynamic brain activity. The fNIRS measures oxygenated hemoglobin (HbO) and deoxygenated hemoglobin (HbR) changes occurring in the cerebral cortex. It provides both reasonable temporal and spatial resolution, it is relatively insensitive to motion artifacts and it allows for an ecological experimental setting. However, the most relevant limitation of fNIRS resides in its inability to provide structural information about the brain. This drawback has become more relevant in recent years, because the earliest fNIRS instruments with only few channels (source–detector pairs) (e.g., Chance et al., 1993, Kato et al., 1993) have been replaced by those with a greater number of sources and detectors (multichannel fNIRS) that allow for a simultaneous functional investigation of a large part of the brain (e.g., Koizumi et al., 2003, Franceschini et al., 2006, Schecklmann et al., 2007; for a review, see Gibson et al., 2005). The latter approach clearly requires more stringent methods for cranio-cerebral correlation, in order to perform a reliable comparison of optical imaging data with the results of other neuroimaging techniques.
Since the proposal by the ICBM (International Consortium for Brain Mapping) of a functional probabilistic atlas (Mazziotta et al., 2000), there has been a constantly increasing use of a standard coordinate space for presenting the data obtained with tomographic functional brain mapping methods, such as functional magnetic resonance imaging (fMRI) and positron emission tomography (PET). Usually, a normalization procedure is performed (by linear and/or nonlinear transformation) to fit individual functional imaging data into a common stereotaxic space referring to a template brain; two brain atlases are commonly used to normalize individual data to a common standard coordinate space: the Talairach (TAL, Talairach and Tournoux, 1988) and the Montreal Neurological Institute (MNI, Collins et al., 1994). While the Talairach atlas is based on a single brain, the MNI defined a new standard brain by averaging several magnetic resonance (MR) scans of human subjects (Collins et al., 1994; for a review, see Brett et al., 2002). The vast majority of functional neuroimaging data is currently presented in MNI space because it is a voxel-based, probabilistic template of the human brain (Mazziotta et al., 2001).
The current standard MNI template, known as the ICBM152 (Mazziotta et al., 2001), was obtained by averaging the high resolution scans of 152 normal subjects. Each MR scan (256 × 256 with 1 mm slices) was normalized to the MNI space using a 9 parameter affine transform. The final resolution of the ICBM152 template is 181 × 217 × 181 with 1 mm isotropic voxels. According to Okamoto et al. (2004), fNIRS data should be presented in standard MNI coordinates to facilitate inter-study and cross-modal comparisons of functional neuroimaging data.
Therefore, it appears fundamental to obtain (directly or indirectly) information on the structural anatomy of the brain which is functionally investigated with fNIRS and to compare reliably the fNIRS results with those coming from other techniques. The direct solution consists in examining the cranio-cerebral structural correspondence by using the MR scan of each subject (or a subset of the subject sample). Although this method provides the highest precision possible, it is time consuming and most importantly, it makes the fNIRS dependent on MRI, decreasing its intrinsic value. MRI scanning is expensive and a scanner might not be available on the same premises of the fNIRS laboratory. As noted by Singh et al. (2005), opting out an fNIRS investigation because of the unavailability of individual MR scans would cause a conspicuous loss of valuable neuroimaging data.
The indirect method to establish cranio-cerebral correlation consists in using the international 10-20 system (Jasper, 1958) or its extended version, the 10-10 system (Chatrian et al., 1985). This is considered the de facto standard in electroencephalography (EEG), because it is a cheap and reproducible method for electrode placement. Essentially, this system describes the scalp of the subject through a series of locations obtained by measuring the distances between cranial landmarks. The main assumption behind the adoption of the 10-20 and 10-10 systems is that there is a systematic correspondence between head surface locations and the underlying cortical regions. Although prior studies investigated cranio-cerebral correlation (Blume et al., 1974, Morris et al., 1986, Homan et al., 1987), the study of Okamoto et al. (2004) was the first to provide a quantitative measure of such correspondence. The authors collected the MR scan of 17 subjects after having marked their 10-20 locations on the scalp, in order to identify them during MRI inspection. They then projected the 10-20 reference points at the cortical level, obtaining a probabilistic distribution of the 10-20 locations in MNI coordinates. Interestingly, they found that 10-20 reference points can be used to estimate the correspondent MNI coordinates (both for the head surface and the cortical projections) with an error that was less than 1 cm (except for the occipital regions, O1 and O2, that exhibited the greatest individual variation, tending to be less reliable).
The seminal study of Okamoto et al. (2004) prompted several investigators to use and refine the cranio-cerebral correlation method (Okamoto & Dan, 2005, Singh et al., 2005, Jurcak et al., 2007, Tsuzuki et al., 2007, Koessler et al., 2009, Custo et al., 2010). Most notably, Tsuzuki et al. (2007) proposed a virtual registration method of fNIRS data onto MNI space that involved neither scanning the subjects with MRI nor digitizing their head surface. They used flexible holders and tested them on human heads and in a virtual environment. Placement of a probe holder on the scalp with the method of Tsuzuki et al. (2007) requires running a simulation of the holder's deformation and registering the position of probes and channels. That is, a virtual holder deformation algorithm is used to mimic the deformation of the holder on the scalp. A set of synthetic heads and brains is generated by randomly combining head sizes and shapes from the un-normalized scans of the NFRI_R17 (National Food Research Institute Reference database, 17 MR individual scans; for details, see Okamoto et al., 2004). For each synthetic head and brain, the virtual holder deformation algorithm is used to estimate the position of fNIRS head surface points of the probes and the cerebral projections of the channels. The data on the position of probes and channels obtained from the synthetic head and brain are transformed to the original MR dataset, and then to MNI space (for details, see Okamoto and Dan, 2005). This procedure is repeated 1000 times, in order to simulate head size and shape variability across population; afterwards, statistical analysis of the MNI coordinates is performed to estimate the most likely location of each channel and its variability. Estimated locations are then anatomically labeled by using conventional brain atlases. Remarkably, they showed that the spatial error implied by different head sizes and shapes of the subjects can be minimized by the use of flexible/elastic probes. Indeed, they showed that the precision of their method was comparable to that of a previous probabilistic registration method performed with a 3D-digitizer (Singh et al., 2005). However, as the authors themselves noted, the use of a simulated dataset to assess the inter-subject variability and the procedure described in their study is not completely straightforward, since it requires the experimenter to provide different parameters and to perform several adjustments for each virtual holder registration.
Our aim was to create a method that combined the operative advantages of Tsuzuki et al.'s (2007) own method with a greatly improved usability. The core of our proposal stems from the fact that the ICBM152 template is the most common stereotaxic platform for tomographic functional brain mapping methods (Mazziotta et al., 2001; for a review, see Brett et al., 2002), since it is the current standard MNI template. If the neuroimaging community has adopted the “brain” represented in the ICBM152 template as the common reference brain, its head surface should be the best candidate to become the common reference head surface. From this point of view, the introduction of a physical model of the ICBM152 head surface could provide several benefits in the probe placement process. For instance, it could help to avoid using a virtual holder deformation algorithm, because the deformation of the holder would be accounted by its placement on the physical model itself; furthermore, it could provide a direct link between the virtual space of MNI stereotaxic coordinates (and the ICBM152 template) and the physical space of fNIRS channel positions. Classical probe location estimation methods (e.g., Tsuzuki et al., 2007) provide an estimate of the MNI coordinates corresponding to the channels only after the probe has been placed. In contrast, with the adoption of a physical model of ICBM152 head surface, experimenters could use the MNI coordinates of the regions of interest to directly guide the probe placement process on the physical model.
Therefore, we propose a novel probe placement method explicitly designed for functional group analysis in multichannel fNIRS, based on a physical model of the ICBM152 template. The presented method pursues three main purposes. First, it must be a fast, simple, reproducible and straightforward method to execute a reasonably precise probe placement. Second, it must be able to avoid the digitizing procedure for every subject and to eliminate the need of the MR of individual subjects, thus making fNIRS group analysis MRI-free. Finally, it needs to ensure an optimal compatibility of the fNIRS results with other neuroimaging results that adopt the MNI coordinates to locate the cerebral regions. Accordingly, our method allows to register fNIRS optode and channel positions directly to the ICBM152 template, and therefore to the MNI stereotaxic coordinate system.
The paper is organized as follows. First, we describe how to create a physical model of the head corresponding to the ICBM152 brain template. We then describe how the physical model is used to find the best placement of the fNIRS probes in relation to the cerebral regions to be investigated. Finally, we provide a practical example of setting up an fNIRS bilateral recording from the parietal cortex, which is also used as an additional validation of the procedure.
Section snippets
Physical model of the ICBM152 head
In order to create a physical model of the head corresponding to the ICBM152 brain template, we executed a series of operations, mostly performed with 3D-DOCTOR ver. 3.5 (ABLE SOFTWARE CORP., http://www.3d-doctor.com), an advanced 3D modeling, image processing and measurement software. The MNC file of ICBM152 template, which is freely available (http://packages.bic.mni.mcgill.ca/tgz/mni-models_icbm152-lin-1.0.tar.gz) at 1 mm3 resolution, has been converted in ANALYZE format using LONI Debabeler,
Discussion
The present method has three main features: (i) it is a fast, simple and straightforward method to obtain a precise probe placement for fNIRS recording; (ii) it does not require individual MRI scans, making fNIRS group analysis independent from individual MR scans, and it does not require the digitizing procedure for every subject; (iii) probes and channels are located in MNI coordinates in a straightforward way, ensuring that fNIRS results can be readily compared with other neuroimaging data.
Acknowledgments
This study was supported by a grant from Cariparo Foundation (Progetti di Eccellenza 2007) to M.Z. We thank D. Basso and L. Bardi for their help with the neuronavigation software and S. Sighel for help with the illustrations.
References (35)
- et al.
Anatomic correlates of the ten-twenty electrode placement system in infants
Electroencephalogr. Clin. Neurophysiol.
(1974) - et al.
Anatomical atlas-guided diffuse optical tomography of brain activation
Neuroimage
(2010) - et al.
Selective activation of the superior frontal gyrus in task-switching: an event-related fNIRS study
Neuroimage
(2008) - et al.
Cerebral location of international 10–20 system electrode placement
Electroencephalogr. Clin. Neurophysiol.
(1987) - et al.
Voxel-based cortical thickness measurements in MRI
Neuroimage
(2008) - et al.
10/20, 10/10 and 10/5 systems revisited: their validity as relative head-surface-based positioning systems
Neuroimage
(2007) - et al.
Automated cortical projection of EEG sensors: anatomical correlation via the international 10-10 system
Neuroimage
(2009) - et al.
A probabilistic atlas and reference system for the human brain: International Consortium for Brain Mapping (ICBM)
Philos. Trans. R. Soc. Lond. B Biol. Sci.
(2001) - et al.
Automated cortical projection of head-surface locations for transcranial functional brain mapping
Neuroimage
(2005) - et al.
Three-dimensional probabilistic anatomical cranio-cerebral correlation via the international 10-20 system oriented for transcranial functional brain mapping
Neuroimage
(2004)
Absorption spectroscopy in tissue-simulating materials: a theoretical and experimental study of photon paths
Appl. Opt.
Spatial registration of multichannel multi-subject fNIRS data to MNI space without MRI
Neuroimage
Co-Planar Stereotaxic Atlas of the Human Brain: 3-Dimensional Proportional System: An Approach to Cerebral Imaging
Automated anatomical labeling of activations in SPM using a macroscopic anatomical parcellation of the MNI MRI single-subject brain
Neuroimage
Near infrared spectroscopy (NIRS): a new tool to study hemodynamic changes during activation of brain function in human adults
Neurosci. Lett.
Retinotopic mapping of adult human visual cortex with high-density diffuse optical tomography
Proc. Natl. Acad. Sci. USA
The problem of functional localization in the human brain
Nat. Rev. Neurosci.
Cited by (89)
Effects of different exercise patterns on drug craving in female methamphetamine-dependent patients: Evidence from behavior and fNIRS
2023, Mental Health and Physical Activity