Animal Care and Use Committee approval
The protocol of the present study was approved by the Institutional Animal Care and Use Committee of the National Institute of Advanced Industrial Science and Technology, Japan, and was carried out in accordance with the guidelines within the “Guide for the Care and Use of Laboratory animals” (Eighth ed., National Academy of Sciences).
Optimization of fNIRS optode distance in monkey’s motor activity measurement
The fNIRS optode arrangement was customized for the present monkey study. Because the superficial layers such as the scalp and skull of the monkey head are thinner and the brain is smaller than those of humans, the optimal source–detector distance for measuring fNIRS signals from the monkey brain was presupposed to be smaller. We estimated the source–detector distance optimal for monkey brains through a calculation simulating light propagation in an optical model of the monkey head.
For this simulation, T1 and T2 weighted magnetic resonance (MR) head images of three macaque monkeys including the trained subject were used and optical models of these heads were constructed. Based on simulations over the three individual models, universality of the calculation results were examined. Each model consisted of six optical layers: air, skull, cerebrospinal fluid, gray matter, white matter, and other soft tissues. T1 and T2 weighted MR images present sufficient contrasts among these layers to identify each23. In the MR image of the trained (scalp-incised) subject after forming optode sockets, the skull surface was not clear because the skull tissue and the socket material (acryl resin) contain few protons and thus have little T1 resonance. In this case, the skull surface was presumed to be identical to the boundary between the scalp and skull in the image before forming the sockets (i.e., the image of the intact subject’s head). The position of this boundary in the post-implantation image was identified by matching the before and after brain tissue boundaries with each other. Overlapping was executed through a rigid body transformation (FSL 5.0, Oxford, UK). The boundaries of other tissues and the optode positions were directly identified using the image after forming the sockets. In cases of other two monkeys, six-layered optical models including soft tissue layer were firstly constructed based on MR images of intact heads. Thereafter, for uniforming layered structures in these models with the scalp-incised model, the superficial soft tissue layer of parietal region was replaced by air layer. The coefficients of absorption and reduced scattering in each tissue at 800 nm wavelength were taken from literature24,25,26,27. A refractive index of n = 1.40 was used for all tissue layers. Light propagation in the models was calculated using the diffusion equation. The finite element method was conducted to solve the diffusion equation28. In the MR images of each monkey, the center position of the hand knob, known as the hand motor center in the M1 area was identified manually by a researcher well-versed in monkey brain anatomy. The position of the hand knob in the optical model was projected onto the scalp surface. Source–detector pairs separated by 5 to 25 mm (in 5 mm steps) were introduced to the model so that the midpoint of each source–detector pair aligned with the projected point (i.e., the hand knob). Optode pairs aligned both parallel and orthogonal to the central sulcus adjacent to the hand knob were examined. In each condition, the detected light intensity, the mean optical path length29, and spatial sensitivity profile (SSP)30 were calculated. The SSP over all the voxels in the model was obtained through the calculation of the photon measurement density function in each voxel and was normalized by mean optical path length. The photon measurement density function provides the probability of light transit through the voxel in a given source–detector condition and was calculated on the basis of the reciprocity principle14,31. Because the detected light travels through each voxel with different sensitivity, the sum of the SSP of voxels within a certain layer is equivalent to the partial optical path length in that layer. For example, in this calculation, the partial optical path length in the gray matter (Lgray) was calculated as the sum of SSP in the corresponding gray matter. To evaluate the effective signal sensitivity for cerebral motor activity, a regional Lgray in a cubic volume of 5.4 × 5.4 × 5.4 mm3 at the hand knob was also calculated. Further details in the simulation calculation are described elsewhere32.
Results of the calculation are shown in Fig. 1. The SSP on the gray matter surface became broader as the source–detector distance (dSD) increased (Fig. 1a), which indicated that shorter distance provided higher spatial resolution but lower detection sensitivity. Similar dependencies of SSP on dSD were observed in all subjects and parallel and orthogonal optode alignment conditions. The predicted detected intensity (I) monotonically decreased with increase in dSD (Fig. 1b) and Lgray monotonically increased (Fig. 1c). However, the regional Lgray at hand knob asymptotically approached to a finite length (Fig. 1d). The predicted values in each calculation were very similar among the cases of subjects and optode alignment conditions. In Fig. 1d, the regional Lgray showed little increase in the range of dSD longer than 15 mm. This indicates that the signal sensitivity for the gray matter in the hand knob region is not gained by increasing dSD more than 15 mm. On the other hand, the Lgray in all over the inter-optode region further increased in this range (Fig. 1c). This indicates the increase in signal contribution from regions other than hand knob, namely, the degradation in spatial resolution of the signal detection in the range more than 20 mm. By taking consideration of these results, we concluded that 15 mm is most appropriate for dSD to detect activity changes in the macaque motor cortex, in which motor representations of the face, hand and arm are located with difference of several millimeters in M133.
One healthy adult Japanese macaque monkey (female; 5.0 kg) without any history of experimentation was used. The positions of M1 and the premotor area (PMA) were determined using stereotaxic coordinates from MR images of the monkey’s brain using a 3.0 T MR imaging (MRI) system (Philips Ingenia 3.0 T, Philips Healthcare, Best, The Netherlands). The anatomical MRI protocols consisted of a T1-weighted turbo field echo sequence (repetition time/echo time, 7.3/3.2 ms; number of excitations, 2; flip angle, 8°; field of view, 134 × 134 mm; matrix, 224 × 224; slice thickness, 0.6 mm; number of slices, 200). Pentobarbital anesthesia was administered at 20 mg/kg, after which the parietal region of scalp was incised and optode sockets were formed on the skull surface with self-curing acrylic resin (UNIFAST II Clear, GC Corporation, Tokyo, Japan). Titanium oxide (KA-30, Titan Kogyo, Ltd., Ube, Japan) was mixed into the resin at a weight ratio of 1:450 to match its optical scattering property to that of the skull. These procedures were done under sterile conditions.
Optode arrangement at parietal region of the monkey head
For discriminating the motor representation in M1 with fNIRS signals, the spatial interval between channels in fNIRS should be several millimeters to discriminate the hemodynamic responses at different gyri, whereas the source–detector distance of 15 mm is required for appropriately detecting gray matter responses. To satisfy both conditions, optodes (about 10 mm in diameter) placed conventionally would have to be very densely packed. We therefore adopted a triangular bidirectional optode arrangement, in which the optodes are placed at regular triangle lattice points, and each is used as a source and a detector through temporal switching16. The schematic illustration of the arrangement is shown in Fig. 2a (Top view). As determined above, the source–detector distance was fixed to be 15 mm; thus, the spatial interval among adjacent channels was 7.5 mm. Each optode has bifurcated ends connected to a source or a detector in the fNIRS OMM-3000 system (Shimadzu Corporation, Japan). The bifurcated optodes were custom made of optical fiber bundles (Moritex Corporation, Japan). Using ternary optodes as one source and two detectors, signals from two channels are collected simultaneously, and by completing illumination at all optodes, two measurements for every channel are accomplished.
As shown in Fig. 2a (Side view), fifteen optodes held with a custom holder were fixed into the previously implanted socket wells on the skull surface just before every experimental session. There, adjacent optodes were cross-linked by two-layered linkage plates. The lower plates hold the distances between optodes as 15 mm. The upper plates were equipped with elongated holes and nuts. After the optode tips of globular shape were inset into the sockets, the optode arrangement was consolidated with locking nuts. Switching between illumination and detection in each optode was controlled through software built into the OMM-3000. Twenty-seven channels were available for measurement in the parietal region of the monkey head. Optical absorbance data at wavelengths of 780, 805, and 830 nm for each channel and the digital signal from the Digital Laser Sensor were recorded by the OMM-3000.