Neural correlates of head restraint: Unsolicited neuronal activation and dopamine release

To minimize motion-related distortion of reconstructed images, conventional positron emission tomography (PET) measurements of the brain inevitably require a firm and tight head restraint. While such a restraint is now a routine procedure in brain imaging, the physiological and psychological consequences resulting from the restraint have not been elucidated. To address this problem, we developed a restraint-free brain PET system and conducted PET scans under both restrained and non-restrained conditions. We examined whether head restraint during PET scans could alter brain activities such as regional cerebral blood flow (rCBF) and dopamine release along with psychological stress related to head restraint. Under both conditions, 20 healthy male participants underwent [15O]H2O and [11C]Raclopride PET scans during working memory tasks with the same PET system. Before, during, and after each PET scan, we measured physiological and psychological stress responses, including the State-Trait Anxiety Inventory (STAI) scores. Analysis of the [15O]H2O-PET data revealed higher rCBF in regions such as the parahippocampus in the restrained condition. We found the binding potential (BPND) of [11C]Raclopride in the putamen was significantly reduced in the restrained condition, which reflects an increase in dopamine release. Moreover, the restraint-induced change in BPND was correlated with a shift in the state anxiety score of the STAI, indicating that less anxiety accompanied smaller dopamine release. These results suggest that the stress from head restraint could cause unsolicited responses in brain physiology and emotional states. The restraint-free imaging system may thus be a key enabling technology for the natural depiction of the mind.

. The restraint-free brain PET system. ( A ) During the PET scans, the participants wore a head-cap with four LED markers. ( B ) The locations of these markers were tracked every four ms by two high-speed cameras (HAMAMATSU Intelligent Vision System, IVS) mounted at the back of the gantry. Using information about head movements monitored during the scans, we corrected the list-mode data with a quaternion algorithm and reconstructed motion-corrected PET images.
with claustrophobia or panic disorder. Indeed, many researchers have pointed out that the use of these imaging devices often invokes both physiological and psychological stress responses ( Tessner et al., 2006 ;Lueken et al., 2011 ;Muehlhan et al., 2011Muehlhan et al., , 2012, and these could affect task-related brain activations ( Keulers et al., 2015 ;Muehlhan et al., 2013 ;Weldon et al., 2015 ). However, the exact mechanisms underlying these stress responses are currently unknown.
Here, we hypothesize that brain activations associated with these physiological and psychological stress responses are partly due to the inevitable body restraint associated with the imaging acquisition. The body restraint is required for minimizing body movements during scanning, as body movements result in a motion-related distortion of reconstructed images. In a previous study using a cardiac-torso phantom, respiration with 11-mm diaphragm motion could cause a mean lesion maximum standard uptake value (SUV max ) underestimation of 28%, and a mean lesion volume overestimation of 130% in PET/CT images with 1 cm-wide lesions ( Liu et al., 2009 ). Head movements during MRI scan acquisition, which are common in psychiatric patients, cause a significant reduction in estimated regional gray matter volumes ( Blumenthal et al., 2002 ) and cortical thickness ( Pardoe et al., 2016 ), and also cause confounding or erroneous interpretation of resting state functional MRI (fMRI) data ( Bolton et al., 2019 ;Makowski et al., 2018 ;Power et al., 2012 ). To minimize such undesirable outcomes, the head restraint is considered a prerequisite for brain imaging studies ( Varrone et al., 2009 ), and various motion correction (MC) algorithms have been developed.
The hypothesis supporting an association between body restraint and brain activation is plausible, given that head restraint or immobilization is one of the most popular experimental paradigms used to investigate stress-related responses in animal research ( Glavin et al., 1994 ;Paré and Glavin, 1986 ). To test this hypothesis, we developed a restraint-free brain PET system ( Fig. 1 ) ( Yoshikawa et al., 2015 ) and conducted PET scans under both restraint and non-restraint conditions. To date, many researchers have proposed various MC techniques for restraint-free neuroimaging data acquisition ( Maclaren et al., 2012 ;Noonan et al., 2015 ;Picard and Thompson, 1997 ;Slipsager et al., 2019 ;Zaitsev et al., 2006 ). In the present study, we adopted the optical marker tracking method for MC, as it offers head motion measurement with high precision and a detailed time scale ( Maclaren et al., 2013 ).
We examined whether head restraint during PET scans could produce altered brain activities in regional cerebral blood flow (rCBF) and dopamine release secondary to stress. We conducted both [ 15 O]H 2 O and [ 11 C]Raclopride measurements, with and without the head restraint. Previous neuroimaging studies have shown that acute stress or the pre-sentation of emotionally negative stimuli increases rCBF in the limbic areas and other cortical regions ( Fusar-Poli et al., 2009 ;Smith et al., 2004 ;Stock et al., 2014 ), which can be observed via [ 15 O]H 2 O measurements. Acute stress has also been associated with a reduction of non-displaceable binding potential (BP ND ) of [ 11 C]Raclopride in the striatum ( Pruessner et al., 2004 ;Tsukada et al., 2011 ), which reflects a phasic increase of extracellular dopamine concentration in this region ( Endres et al., 1997 ;Grace et al., 2007 ). We predicted that these brain responses should be observed in the restraint condition, due to higher stress.
The protocol we used for [ 15 O]H 2 O and [ 11 C]Raclopride PET measurements was similar to that of standard PET measurements, other than the existence of the restraint and non-restraint conditions ( Fig. 2 ). During the PET scans, participants solved working memory (WM) tasks using the basic two-back test in both the restraint and non-restraint conditions. Solving this kind of cognitive task in the scanner is a commonly used experimental design in cognitive neuroscience Cohen et al., 1997 ), and we used it to address the effect of head restraint on the neurophysiological responses. Before, during, and after each PET scan, we also measured the physiological stress responses, including plasma stress hormones (cortisol, adrenalin, noradrenalin, dopamine, and adrenocorticotropic hormone [ACTH]), electrocardiogram (ECG), and electrodermal activity (EDA). We also measured subjective and psychological stress responses using the State-Trait Anxiety Inventory (STAI) before the first PET measurement and immediately following each PET session. We then examined whether the rCBF and dopamine release in the brain correlated with these subjective stress responses. The phantom was mounted on a moving stage (SGSP46-300; SIGMA KOKI, Tokyo, Japan), and was situated at the center of the gantry. The LED marker used in human measurements was attached on top of the phantom to track its position. A 10-min static scan was acquired to provide a reference scan free of any movement. Six 10-min scans with movement were then performed, with the movements in translation and rotation. The translation movements were applied ± 30 mm from the original position in either the X or Y direction, at a speed of 10 mm/s, or in the Z direction at a speed of 6 mm/s, during scanning. The skew movements were applied 0°-30°or 0°-60°from the original angle around the Z axis at a speed of 2.7 °/s, and an anteroposterior rotation angle of ± 10°around the X axis at a speed of 0.7 °/s during scanning. A 60-min transmission scan was performed 15 h after the emission scans. The PET images were reconstructed with the 3D list-mode dynamic row-action maximum likelihood algorithm (DRAMA) ( Nakayama and Kudo, 2005 ) on a matrix of 128 × 128 × 42 voxels (voxel size: 2.6 × 2.6 × 3.4 mm) with or without MC.

Participants
Thirty healthy adult volunteers were recruited. Data from ten participants were excluded from analyses for the following reasons: PET machine scanning trouble ( n = 4), failure in cyclotron tracer synthesis ( n = 3), failure in visual task presentation ( n = 2), and difficulty in tracer injection due to thin blood vessels ( n = 1). A total of 20 participants (age 32.2 ± 10.1; mean ± standard deviation [SD]) were included. All participants were male, to eliminate the potential effects of sex and PET over two days. During the PET scans, participants solved the working memory (WM) task in the PET scanner, with (restraint) or without (non-restraint) head restraint. Before, during, and after each PET scan, we also measured physiological stress responses, including plasma stress hormones, electrocardiogram (ECG), and electrodermal activity (EDA). Before the first session of PET scanning and just after each session, we obtained the psychological stress responses of the participants using the State-Trait Anxiety Inventory (STAI). The order of the restraint and non-restraint conditions was alternated between the [ 15 O]H 2 O and [ 11 C]Raclopride PET experiments, and was counter-balanced among participants. menstrual cycle ( Kirschbaum et al., 1999 ). All participants were righthanded, had a normal or corrected-to-normal vision, were free from psychiatric or neurological abnormalities including claustrophobia, and did not use medication that could influence cognitive functioning or stress responses. Ten participants were scanner-naïve, while the others had experienced PET and/or MRI scans. The study was conducted in compliance with the Declaration of Helsinki and was approved by the institutional review board of the Hamamatsu Medical Photonics Foundation (No. 118) and Hamamatsu Photonics K.K. (H-73), and written informed consent was obtained from all participants.
To determine the sample size, we referenced our prior data of dopamine release ( Fukai et al., 2019 ). Referring to the data, 20 participants were sufficient to be able to reject the null hypothesis with a probability (power) of 0.8 because the true difference in the mean response (significant dopamine release) of matched pairs was 0.15 under the SD 0.22. For the sample size calculation, we used G * Power version 3.1 ( Faul et al., 2007 ).

PET acquisition
We used a brain PET scanner (SHR12000; Hamamatsu Photonics K.K., Hamamatsu, Japan). In the [ 15 O]H 2 O and [ 11 C]Raclopride experiments, participants received two sessions with (restraint) or without (non-restraint) head restraint. Under the restraint condition, the participant's head was firmly restrained with a thermoplastic face mask (ES-FORM; Engineering System Co., Ltd., Nagano, Japan), whereas the mask was removed under the non-restraint condition. Immobilization with a thermoplastic mask can guarantee a more reliable measurement than the use of inflatable bags or tape, especially for patients who have difficulty in refraining their body movement (e.g., patients with dementia and psychiatric disorders). This immobilization technique is often used during MRI or CT scans, especially in the field of radiotherapy, where patient immobilization is critical for successful imaging and treatment, and its use and efficacy have been recently studied ( Chandarana et al., 2018 ;Mandija et al., 2019 ;Wang et al., 2016 ).
Under both conditions, the participants were instructed to lay in a supine position and refrain from falling asleep in the scanner gantry.  ( Peirce, 2007 ), and visual stimuli were presented on a 15-inch monitor (CV512PJ, JVC KENWOOD, Yokohama, Kanagawa, Japan) placed approximately 60 cm in front of the participant. The participants were asked to monitor the identity of a series of one-digit numbers from ''0 ′ ' to ''9 ′ ', presented in a random sequence. They had to push a button with their right thumb using a response box ( OTR-1x4-L , Current Designs, Philadelphia, PA, USA) to indicate if the currently presented stimulus was the same as the one presented two trials previously. The participants received a total of six task blocks for the [ 15 O]H 2 O experiment and 20 task blocks for [ 11 C]Raclopride experiment. The detailed protocol of the WM task is provided in Supplementary Fig. S1 and Supplementary Methods.

State-trait anxiety inventory
To assess the subjective stress of the participants, we used a Japanese version of the STAI ( Spielberger, 1983 ). The participants completed the questionnaires before the first session of each experiment and just after each PET session.

Collection and measurement of plasma stress hormones
Plasma probes for cortisol, dopamine, adrenalin, noradrenalin, and ACTH measurement were sampled from an intravenous catheter in the  (T6); and 20 min after the second session (T7). The data from T1, T4, and T7 were regarded as a baseline.
In the [ 11 C]Raclopride experiment, the time-points were as follows: before entering the PET gantry for the first session (T1); 30 min after entering the scanner gantry and before the onset of the [ 11 C]Raclopride measurement for the first session (T2); 30 min after the onset of [ 11 C]Raclopride measurements (T3); 20 min after the first session (T4); before entering the PET gantry for the second session (T5); 30 min after entering the scanner gantry and before the onset of [ 11 C]Raclopride measurements for the second session (T6); 30 min after the onset of [ 11 C]Raclopride measurements (T7); and 20 min after the second session (T8). The data from T1, T4, T5, and T8 were regarded as a baseline.
All samples were immediately centrifuged and stored at − 20 °C. The samples were anonymized and sent for hormonal measurement to a clinical laboratory testing company (SRL, Inc, Tokyo, Japan), which was blind to the experimental conditions. The cortisol and ACTH plasma concentrations were measured using a commercial electro-chemiluminescence-assay (ECLIA) kit (Cobas R ○ , Roche Diagnostics GmbH, Mannheim, Germany). The dopamine, adrenalin, and noradrenalin plasma concentrations were measured using a company's in-house high-performance liquid chromatography (HPLC) kit. For adrenalin and dopamine, concentrations less than 5 pg/ml were not detectable. Participants with samples under this limit were thus discarded from the analyses. Raclopride experiments, respectively, were above this limit. The data for dopamine were thus excluded from all analyses. To assess the overall secretion over a specific time period, we calculated the area under the curve with respect to the ground (AUC G ), which is outlined in the following formula ( Pruessner et al., 2003 ).
(1) t i denotes the individual time distance between measurements. m i is the value of individual measurement and n is the total number of measures. AUC G takes into account the difference between the single measurements from each other (i.e., the change over time) and the distance of these measures from the ground or zero (i.e., the level at which the changes over time occur). The calculated AUC G for each condition was then used for subsequent statistical analyses.

Electrocardiogram and electrodermal activity measurement
Electrocardiogram and electrodermal activity were continuously recorded during the experimental sessions using a BIOPAC MP 160 device (BIOPAC Systems Inc., Goleta, CA, USA) and AcqKnowledge 5.0 software for data acquisition. The ECG signal was measured using three disposable EL 503 ECG electrodes (Biopac Systems Inc., Goleta, CA, USA) positioned in an Einthoven Lead I configuration and connected to the BIOPAC amplifier module ECG100C. Electrodermal activity was measured using two disposable EL 507 EDA electrodes (Biopac Systems Inc., Goleta, CA, USA) placed on the ring and little fingers of the left hand and connected to the BIOPAC amplifier module EDA100C. All electrodes were attached more than 5 min before the measurements were taken to ensure the gel was sufficiently absorbed by the skin of the measurement area ( Braithwaite et al., 2013 ). All signals were digitized at a rate of 1,000 Hz. Heart rate was extracted from the ECG signals using the peak-detection function implemented in AcqKnowledge (version 5.0). Participants' movement, coughs, sighs, and/or sneezes were tagged as events on the data, and only artifact-free intervals of 1 min during the rest condition and each WM task in the restraint and non-restraint conditions, were used for calculating the AUC G .

PET data reconstruction and analyses
During the PET scans, the participants wore a head-cap with four LED markers. The locations of these markers were tracked every four ms by two high-speed cameras (HAMAMATSU Intelligent Vision System, IVS) mounted at the back of the gantry. The total distance of head movement and the displacement from its original position along six directions are shown in Supplementary Table S1. Using information about head movements monitored during the scans, we corrected the line-ofresponse (LOR) of list-mode data and reconstructed the image using a 3D list-mode dynamic row-action maximum likelihood algorithm (DRAMA) ( Nakayama and Kudo, 2005 ). In this reconstruction process, j th voxel of reconstructed image was obtained with the following iterative processes: k and l represent the number of main and sub iterations as described in the original article of DRAMA, respectively. represents the relaxation coefficient. S represents the subset of list data. a ij is the probability of detection of the coincidence event in voxel j for a LOR i . is the relaxation parameter to control the decay speed of the relaxation coefficient. In the present measurement, we set as 30. L represents the motion of LOR i of n th coincidence in specific time t detected by high-speed cameras. The motion of LOR i was corrected by the inverse of L , which is formulated as follows: M represents the relative motion of LED marker to t = 0, Q represents the quaternion of LED marker rotation, [ ∆x ∆y ∆z ] T represents the shift of LED marker centroid, while [x i 1 y i 1 z i 1 ] T and [x i 2 y i 2 z i 2 ] T represent the three-dimensional positions of both endpoints of LOR i . C is a motion-compensated sensitivity image of PET scanner as defined by m is the motion information index, and ∆t is the sampling pitch of motion information. The measurement of head motion and image reconstruction was accomplished with our in-house program. The accuracy of our MC technique was confirmed using non-human data in a Hoffman phantom study ( Fig. 3 ), and human data from the [ 11 C]Raclopride PET dynamic measurements in normal participants from the present study ( Fig. 4 and Supplementary Fig. S2). Additionally, the accuracy was further verified using PET dynamic data of patients with Alzheimer's disease ( Supplementary Fig. S3  [ 15 O]H 2 O PET volumes. The coregistered structural images were spatially normalized to the standard brain space as defined by the Montreal Neurological Institute (MNI), using the unified segmentation algorithm with very light regularization, which is a generative model that combines tissue segmentation, bias correction, and spatial normalization in the inversion of a single unified model ( Ashburner and Friston, 2005 ). After spatial normalization, the resultant deformation field was applied to the realigned [ 15 O]H 2 O PET imaging data, which was resampled ev-ery 3 mm using seventh-degree B-spline interpolation. All normalized functional images were then smoothed using an isotropic Gaussian kernel of 8 mm full width at half maximum (FWHM). We assumed that the scan-to-scan variability within a PET session and the session-by-contrast interactions were approximately equal in our [ 15 O]H 2 O PET measurements. We thus adopted a fixed-effect analysis with multi-participant and multi-condition full factorial design and included all scans of all participants into a single general linear model of SPM. Condition-specific effects were estimated with a general linear model, whereas the order of PET image acquisition, participants, and global signal intensities were used as covariates in the design matrix to account for the effect of these variables and global signal normalization. The order of PET image acquisition was treated as a covariate so that the influence of the restraint would not be all or nothing, and would be affected by time. A relative threshold of 0.5 was administered to eliminate the signals from outside the brain. The statistical parametric maps were thresholded at a voxel level of uncorrected p < 0.001, and at a cluster level of corrected p < 0.05 using the false discovery rate (FDR).
Using PMOD software (version 3.0, PMOD Technologies Ltd., Zurich, Switzerland), the BP ND of [ 11 C]Raclopride in the putamen was calculated based on a simple reference tissue model with two compartments ( Lammertsma and Hume, 1996 ), in which the cerebellum was selected as the reference region. The ROIs were determined bilaterally from coregistered MRI images from each individual using an MR image atlas ( Mai et al., 2015 ). These ROI values of BP ND were then used for subsequent statistical analyses. We also conducted voxel-wise analyses of BP ND with SPM for supplementary analyses. BP ND parametric maps were normalized using the same protocol as the [ 15 O]H 2 O PET analysis and smoothed with an isotropic Gaussian kernel of 8 mm FWHM. An absolute threshold of 1.0 was administered to eliminate the signals of nonspecific binding regions, and the statistical parametric maps were thresholded at a voxel level of uncorrected p < 0.005, which is the same threshold used in our previous study ( Nozaki et al., 2013 ).

Statistical analyses
Statistical analyses were performed using R (version 3.6.1). The effect of head-restraint on stress responses was determined using mixedeffect analysis of variance (ANOVA) models with the condition (restraint and non-restraint) as a within-participants factor, and order of condition as a covariate to control for a possible order effect ( Lueken et al., 2012 ;Tessner et al., 2006 ). To account for the difference between the two experiments, we conducted mixed-effect ANOVA models with experiment ([ 15 O]H 2 O and [ 11 C]Raclopride measurements) as a within-participants factor. We also examined the correlation between the difference in STAI under the restraint and non-restraint conditions, and that of the BP ND of [ 11 C]Raclopride in the putamen ROI.

Data availability
The datasets generated during and/or analyzed during the current study are not publicly available due to the absence of agreement from the participants but are available from the corresponding authors on reasonable request.

Validation of the MC technique
We first examined the validity of our MC technique with a 3D Hoffman phantom. Our MC technique could successfully cancel out both the translation and skew movements ( Fig. 3 A). The line profiles of these images indicated that the SUVRs of the data with MC were comparable with those with no movement ( Fig. 3 B).
We then examined the effect of MC for the time activity curves (TACs) of the [ 11 C]Raclopride PET experiment ( Fig. 4 and Supplemen- tary Fig. S2). We confirmed that MC could effectively reduce a motionrelated deviation of SUV.

Comparisons of psycho-physiological stress responses between the restrained and non-restraint conditions
We compared the physiological and psychological stress responses, including plasma stress hormones (cortisol, adrenalin, noradrenalin, and ACTH), ECG, EDA, and STAI, under the restrained and non-restraint conditions ( Table 1 and Supplementary Fig. S4). The data regarding dopamine were excluded from the analyses, due to the extremely low concentrations of dopamine observed (See Materials and Methods). Mixed-effect ANOVA models showed no significant effect of head restraint on stress responses ( p > 0.05). These results may be due to the limited number of participants, the existence of poor-responders, and/or the effect of anticipatory stress. We also examined the differences in WM task performance, and found no significant effect of the head restraint. As the 2-back test was relatively easy for participants to complete, as shown by their high scores (244 task blocks out of 520 blocks were completed 100% correctly), this result may be due to a 'ceiling effect'. A mixed-effect ANOVA model used to assess the differences between

rCBF analyses of [ 15 O]H 2 O
To examine whether the increase in rCBF was caused by the stress from the head restraint, we compared [ 15 O]H 2 O responses between the restraint and non-restraint conditions. In the restraint condition, we found significantly higher activation in the right parahippocampus, brain stem, and motor/somatosensory cortices including the supplementary motor area, right precentral gyrus, and bilateral postcentral gyri compared to the non-restraint condition ( Fig. 5 and Supplementary Table S2, FDR corrected p < 0.05). The activated areas in the brain stem included the region around the dorsal nucleus of the vagus nerve, nucleus ambiguus, glossopharyngeal nerve, and nucleus solitarius. The parahippocampus is a part of the limbic area, and its activation is frequently All figures are mean ± SD. Abbreviations: AUC G , area under the curve with respect to the ground; ACTH, adrenocorticotropic hormone; ECG, electrocardiogram; EDA, electrodermal activity; STAI S, state anxiety score of state-trait anxiety inventory; STAI T, trait anxiety score of state-trait anxiety inventory; WM, working memory; d', d-prime; RT, reaction time.
observed in stressful conditions. The observed activation of this region is likely related to the stress induced by head restraint. The activation time courses in each region for all participants are shown in Supplementary Fig. S5. We examined the correlation between rCBF and the total distance of head movement and found no significant relationship.
These results indicate that the observed activation during the restraint condition cannot be explained by the artifact of our MC or head motion itself.
We also conducted whole-brain analyses for the images without MC. This analysis revealed that activation clusters in the supplementary motor area, precentral gyrus, and postcentral gyri were merged together, whereas other clusters were non-significant ( Supplementary Fig. S6 and Supplementary Table S3). These results might be due to blurring in the images without MC, and the large medial activation clusters may not have been as susceptible to motion effects.

Analyses of BP ND of [ 11 C]Raclopride
Time activity curves (TACs) for [ 11 C]Raclopride were obtained from the putamen and cerebellar ROIs. From these TACs, we calculated the BP ND of [ 11 C]Raclopride in the putamen ROI and compared it between the restraint and non-restraint conditions. We found that BP ND in the putamen was significantly reduced in the restraint condition compared to the non-restraint condition ( Fig. 6 , F (1,18) = 5.85, p < 0.05). We also examined the correlation between BP ND and the total distance of head movement and found no significant relationship. This result again indicates that the observed difference in BP ND cannot be explained by the artifact of our MC or head motion itself. We found an increase in dopamine release in the putamen under this condition. The voxel-wise analyses of BP ND showed significant BP ND reduction in the left putamen and caudate during the restraint condition compared to the non-restraint condition ( Supplementary Fig. S7).

Correlation analyses BP ND of [ 11 C]Raclopride and stress responses
We also examined whether the observed restraint-induced change in BP ND was correlated with a shift in the anxiety score of the STAI. Since a difference in [ 11 C]Raclopride BP ND is considered an index of dopamine release ( Ouchi et al., 2002 ), we examined the correlation between the difference in STAI scores under the restraint and non-restraint conditions, and that of the [ 11 C]Raclopride BP ND in the putamen ROI ( Fig. 7 and Supplementary Fig. S8). We found a significantly negative correlation between these variables ( r = − 0.56, p < 0.05), indicating that less anxiety was associated with smaller dopamine release. However, a negative correlation was not observed for the trait anxiety score of STAI ( r = − 0.0067, p = 0.98). These results show that transient anxiety induced by the head restraint modulates dopamine release in the putamen. For supplemental analyses, we also examined the  correlation between [ 11 C]Raclopride BP ND and other stress responses and found no significant correlations among them ( Supplementary  Fig. S9).

Discussion
We examined whether head restraint during PET scans could produce altered brain activities in rCBF and dopamine release related to the resulting stress. O PET data revealed a higher rCBF for the restraint condition in the right parahippocampus; brain stem; and motor/somatosensory cortices including the supplementary motor area, right precentral gyrus, and bilateral postcentral gyri. We also found that the BP ND of [ 11 C]Raclopride in the putamen was significantly reduced in the restraint condition. Moreover, the restraint-induced change in BP ND correlated significantly with the shift of the state anxiety score of STAI, indicating that less anxiety accompanied smaller dopamine release.
Many researchers have noted that neuroimaging data acquisition can invoke both physiological and psychological stress responses ( Tessner et al., 2006 ;Lueken et al., 2011 ;Muehlhan et al., 2011Muehlhan et al., , 2012, although we did not find any significant effect of the head restraint in our current experiment. These results may be due to the limited number of participants, the existence of poor-responders ( Wust et al., 2000 ), and/or the effect of anticipatory stress ( Dantendorfer et al., 1997 ;Gossett et al., 2018 ). Previous studies investigating the time course of salivary cortisol levels during fMRI scanning found that only five out of 39 participants  or five out of 20 participants ( Lueken et al., 2012 ) were responders for the scanning stress. It has also been shown that the stress elicited by the anticipation of fMRI scanning leads to acute elevations in cortisol prior to scanning ( Gossett et al., 2018 ). In both [ 15 O]H 2 O and [ 11 C]Raclopride experiments, the first time-points for stress response were evaluated just before entering the PET gantry for the first session. One possible reason for our results is that some participants at these time points may have had strong anticipatory stress, which may have interfered with stress-induced responses caused by the head restraint during scanning. Alternatively, focusing on the WM task may have effectively dampened anxiety, since participants are usually comfortable in the scanner if they have some task to perform. For STAI, at least two possible explanations can account for the current results. First, the questionnaire was completed before and after the scan, not during the actual stressor, which may have diluted the effects. Second, half of the participants had an experience of PET and/or MRI scans, further weakening the potential effect, as they had prior knowledge of how the imaging feels.
It is possible that the physical MC (head restraint) was simply more effective, yielding a better signal. However, the correlation between rCBF and head movement was not significant, which indicates that the observed activation during the restraint condition cannot be explained by an artifact of our MC or head motion itself. Moreover, the same argument cannot be used for the raclopride data, in which the opposite effect was observed. Further, it is possible that a within-participants comparison with repeat injections could control for the effect of head restraint that we observed. However, whether these effects are canceled out will remain unknown without data from non-restraint measurements. In receptor studies with absolute quantitation, a non-subtraction comparison mode is often used; therefore, the results in such studies could be influenced by these effects.
It should be noted that we only recruited male participants to eliminate the potential effects of sex and menstrual cycle. Thus, there could be a potential sex bias in the results. A thorough investigation including female participants is needed to discern the generalizability of our results. In the future it might also be useful to include other measures of discomfort in addition to anxiety; for example, the experience of pain and discomfort is common when using the thermoplastic masks.
The analyses of the [ 15 O]H 2 O PET data revealed a higher rCBF in the right parahippocampus during the restraint condition. The parahippocampus has anatomical and functional connections to the other emotion-related limbic and cortical regions, including the amygdala, cingulate cortex, and insular cortex ( Kemppainen et al., 2002 ;Roy et al., 2009 ;Stefanacci et al., 1996 ;Suzuki and Amaral, 1994 ). Previous neuroimaging studies have shown that acute stress or presentation of emotionally negative stimuli induces a higher rCBF in the limbic areas, including the parahippocampus ( Fusar-Poli et al., 2009 ;Smith et al., 2004 ;Stock et al., 2014 ), and that patients with a lesion of the parahippocampus exhibited impaired recognition of unpleasant music with dissonance ( Gosselin et al., 2006 ). These findings are consistent with our results, based on discomfort by the head restraint causing activation of this region. Some researchers have proposed that the parahippocampus is also associated with contextual processing rather than having a central role in emotional processing and that this region mediates the connection between contextual processing and emotion, facilitating the understanding of emotions and the expectations of the environment ( Aminoff et al., 2013 ). This hypothesis also fits our results, as the participants in the restraint condition would experience and evaluate an unfamiliar laboratory environment with an unpleasant head restraint.
The activated area in the brain stem included the region around the dorsal nucleus of the vagus nerve, nucleus ambiguus, glossopharyngeal nerve, and nucleus solitarius. It has been shown that vagal afferents terminating in the nucleus of the solitary tract are activated by adrenaline with -adrenoceptors, and that noradrenergic cell groups in this brain region send direct projections to the amygdala, which contributes to stress-induced brain activation ( Roozendaal et al., 2009 ). A previous fMRI study showed that inflammatory responses induced by laboratorybased social stress were correlated with activation of the brain area around this region, as well as the parahippocampus ( Slavich et al., 2010 ). We also found higher rCBF in the motor/somatosensory cortices, including the supplementary motor area, right precentral gyrus, and bilateral postcentral gyri. The supplementary motor area has been associated with the planning and execution of motor responses. Considering that participants' behavioral responses are greatly restricted in the restraint condition, the activation found in these motor-related regions could be related to a burden of hand movement for solving the WM task under this condition. Another explanation may be an augmentation of the sensation of somatosensory stimuli in this condition. The superior part of the postcentral gyrus and sulcus is known to be activated by tactile stimulation of the face ( Huang et al., 2012 ;Sereno and Huang, 2006 ). The activation found in the present study could be caused by tactile stimulation of the face by the head restraint. Considering these results together, it is clear that head restraint can cause superfluous rCBF responses in the limbic and sensorimotor brain regions.
We also found that BP ND of [ 11 C]Raclopride in the putamen was significantly reduced in the restraint condition, which reflects an increase in dopamine release under this condition. This result is consistent with previous studies, which reported a lower BP ND of [ 11 C]Raclopride in the striatum that was related to stress ( Pruessner et al., 2004 ;Tsukada et al., 2011 ). Although this region is often associated with motor processing, previous studies have shown that the BP ND of this region is modulated by various emotional or motivational factors, including an expectation of therapeutic benefit induced by placebo ( de la Fuente-Fernández et al., 2001 ), or emotional arousal during music listening ( Salimpoor et al., 2011 ). A meta-analysis of PET and fMRI literature revealed that activations in the basal ganglia, including the ventral striatum and putamen, have been observed in 70% of participants in relation to positive emotions of happiness, and 60% of the participants in relation to the negative emotion of disgust ( Phan et al., 2002 ). The neurodegenerative diseases affecting the striatum, such as Huntington's disease ( Robotham et al., 2011 ;Snowden et al., 2008 ) and Parkinson's disease ( Kan et al., 2002 ), are known to cause significant impairments not only in motor processing but also in recognition of emotion. Together with our finding that showed a correlation between the restraint-induced change in BP ND and the shift of the state anxiety score of STAI, these results indicate that dopamine release in this brain region is modulated by the subjective stress induced by head restraint.

Conclusions
In the present study, we have shown that the stress from head restraint could cause unsolicited responses in brain physiology and emotional states. The results also show that the amplitude of this brain response was correlated with subjective stress related to head restraint. This emphasizes the need for careful examination in interpreting neuroimaging data in relation to head restraint. The restraint-free imaging system could become a key enabling technology for a successful depiction of a natural state of mind.

Declaration of Competing Interest
T.I., M.I., M.F., K.S., S.I., T.S., A.K., H.O., and E.Y. are research employees of Hamamatsu Photonics K.K. The company had no control over the interpretation, writing, or publication of this work.