One of the greatest challenges of modern neuroscience is to discover the neural mechanisms of consciousness and to explain how they produce the conscious state. We sought the underlying neural substrate of human consciousness by manipulating the level of consciousness in volunteers with anesthetic agents and visualizing the resultant changes in brain activity using regional cerebral blood flow imaging with positron emission tomography. Study design and methodology were chosen to dissociate the state-related changes in consciousness from the effects of the anesthetic drugs. We found the emergence of consciousness, as assessed with a motor response to a spoken command, to be associated with the activation of a core network involving subcortical and limbic regions that become functionally coupled with parts of frontal and inferior parietal cortices upon awakening from unconsciousness. The neural core of consciousness thus involves forebrain arousal acting to link motor intentions originating in posterior sensory integration regions with motor action control arising in more anterior brain regions. These findings reveal the clearest picture yet of the minimal neural correlates required for a conscious state to emerge.
Consciousness is a central question in neuroscience (Crick and Koch, 2003). Yet, the neural correlates of consciousness remain poorly understood. Consciousness research can roughly be divided into two categories: (1) studies on the contents of consciousness and (2) studies on consciousness as a state (Dehaene and Changeux, 2011). Whereas theories on the particular contents of consciousness, such as visual consciousness, argue for the importance of cortical structures (Lamme, 2006), theories focusing on consciousness as a state stress the importance of subcortical (Baars, 1995; Fuller et al., 2011) or thalamocortical structures (Tononi, 2004; Merker, 2007). Having awareness of the environment or of one's self is fundamentally based on being in a conscious state (Fig. 1). There is limited human data on which brain structures engage to serve this foundation of consciousness. This study was designed to reveal the minimal neural correlates associated with a conscious state.
Anesthesia is a tool for studying consciousness as it allows the level of consciousness to be directly manipulated (Beecher, 1947; Alkire et al., 2008; Stamatakis et al., 2010; Schrouff et al., 2011; Xie et al., 2011). However, use of this tool requires overcoming the limitation that a drug-induced unconscious state is necessarily confounded by the presence of the drug used to induce unconsciousness (Tononi and Koch, 2008). To illustrate, anesthesia may reduce neuronal activity throughout the brain, yet only its actions in select areas might be directly related to its ability to induce unconsciousness (Franks, 2008). Also, due to differences in hypnotic mechanisms among anesthetic agents (Alkire et al., 2008; Brown et al., 2010), studying only one drug will make generalizations problematic.
In the present study, we overcame these limitations in three distinct ways. First, we used neuroimaging in conjunction with pattern analysis methodology to dissociate the state-related changes in consciousness from the global effects of anesthesia (McIntosh and Lobaugh, 2004). Second, we eliminated the drug-dose change effect on consciousness by imaging a rapid return to consciousness from the unconscious state induced by the unique anesthetic agent dexmedetomidine, a selective α-2-adrenergic agonist, which allows awakening during constant dosing of the drug (Maksimow et al., 2007). Third, with neuroimaging we dissociated the drug-specific effects on consciousness by showing similar activation to arousal from anesthesia with two different drugs. With these approaches we were able to reveal the minimal neural correlates for arousal into a conscious state.
The ability to rapidly and repeatedly change a person's state of consciousness with anesthesia is limited by the pharmacodynamics of anesthetic drugs. Whereas people can be quickly rendered unconscious with anesthesia, the emergence from anesthesia is often variable (Friedman et al., 2010). Thus, imaging techniques, such as functional magnetic resonance imaging, that must overcome their inherent signal-to-noise problems by using multiple behavioral transitions occurring in a relatively short time frame will have limited utility for imaging the changes in consciousness caused by anesthesia. Here, we used positron emission tomography (PET) to image the brain activity changes associated with varying states of consciousness.
Materials and Methods
After obtaining written informed consents and the approvals of the Ethical Committee of the Hospital District of Southwest Finland (Turku, Finland) and the National Agency for Medicines, 20 healthy, right-handed male volunteers (19–28 years) participated in this open, nonrandomized study. All subjects abstained from alcohol use or medication for 48 h and fasted overnight before the study sessions. No premedication was given.
A right forearm vein was cannulated for administration of the anesthetic agents and [15O]H2O. Ringer's acetate infusion was used to keep the intravenous catheter open. The left radial artery was cannulated for arterial blood activity measurements. A GE Datex-Ohmeda S/5 anesthesia monitor (GE Healthcare) was used to record subjects' noninvasive blood pressure, oxygen saturation, electrocardiogram, and the inspiratory and expiratory gas levels of oxygen and carbon dioxide throughout the study. Bispectral index and spectral entropy values were recorded with disposable electroencephalograph (EEG) electrode strips placed on each subject's forehead; the results of which have been presented previously (Kaskinoro et al., 2011).
General anesthesia is often defined as comprising pharmacologically induced unconsciousness (loss of the ability to feel or experience anything), amnesia (forgetfulness), analgesia (lack of pain), and immobility (not moving in response to surgical stimulation) (Brown et al., 2010). In this study, without surgical stimulation, the term “anesthesia” is used to describe pharmacologically induced loss of consciousness (LOC). Its objective behavioral criterion was unresponsiveness to a verbal command.
Subjects were divided into two groups to receive either dexmedetomidine (n = 10) or propofol (n = 10) via computer-driven target-controlled infusion (TCI). Due to wide individual variability in the anesthetic dose required for LOC, scanning subjects on a fixed dose of an agent would place them in different behavioral states. To account for this, each subject's individual concentration for LOC was first determined in a separate study session (Session 1) one month prior to the neuroimaging (Session 2). The individual doses determined in Session 1 were then used to guide drug dosing during neuroimaging in Session 2. Once a subject was assigned for a particular anesthetic agent in Session 1, that subject then again received that anesthetic during neuroimaging in Session 2.
Dexmedetomidine (Precedex 100 μg/ml; Orion).
This was administered via TCI. A Harvard Model 22 syringe pump (Harvard Apparatus) connected to a portable computer running STANPUMP software (by Steven L. Schafer, MD, available at http://www.opentci.org/doku.php?id=code:code) and the pharmacokinetic parameters of Talke (2003) were used. In Session 1, the infusion was started at a target (plasma) concentration of 1.0 ng/ml, followed first by a 0.5 ng/ml target concentration increase, and then by subsequent 0.25 ng/ml increases thereafter (i.e., 1.0 to 1.5, 1.5 to 1.75, 1.75 to 2.0, and 2.0 to 2.25 ng/ml) until LOC was achieved.
Propofol Lipuro 10 mg/ml (B. Braun) was administered via TCI using the Marsh pharmacokinetic model (Marsh et al., 1991). In Session 1, the infusion was started at a target (plasma) concentration of 1.0 μg/ml, followed first by a 0.5 μg/ml target concentration increase, and then subsequent 0.25 μg/ml increases thereafter until LOC was achieved.
In Session 1, infusions were started at a low rate and then increased in a stepwise manner at 10 min intervals until LOC. Consciousness was tested at 5 min intervals by asking the subjects to open their eyes. LOC was defined as being unresponsive to this request. In Session 2, neuroimaging data were obtained during four different constant states of consciousness induced with either dexmedetomidine or propofol. Subjects were scanned (1) at baseline (BL), (2) during sedation (Sed), (3) upon LOC, and (4) upon the return of consciousness (ROC) during the emergence from the anesthesia. To achieve these behavioral states, the drug infusions were aimed at 50, 75, and 100% of the individually predetermined doses for LOC. Consciousness was tested once during each concentration level, i.e., 8 min after the infusion was started or increased. Scanning was started 2 min after the consciousness testing, and no testing was performed during neuroimaging. If the LOC concentration defined in Session 1 was insufficient to produce LOC in Session 2, the drug infusion was increased an additional 25% and scanning was repeated. Infusion rates were adjusted upward at 12 min intervals until the LOC condition was reached and imaged. A sedative state was defined as the last testing condition before LOC where subjects could still open their eyes. During both sessions, subjects were instructed to keep their eyes closed unless specifically asked to open them.
After the LOC scan was completed, consciousness for those in the dexmedetomidine group was restored using gentle tactile and verbal stimulation while the subjects were still receiving an uninterrupted continuous steady drug infusion. After subjects regained their consciousness, a scan for ROC condition was performed. Subsequently, an additional scan (LOC-2) was performed when these subjects again lost consciousness after they were left unstimulated in the scanner for a few minutes. For those in the propofol group, the infusion was terminated and the subjects were allowed to emerge from the anesthesia. Consciousness was tested at 1 min intervals until ROC. When a conscious state had returned, the subjects were scanned in the ROC condition. ROC was defined as the first eye opening in response to command after LOC.
A previously described PET procedure for measuring regional cerebral blood flow (rCBF) was used (Kaisti et al., 2002). Scans were obtained with an ECAT EXACT HR+ PET scanner (Siemens/CTI) and [15O]H2O (300 MBq) was used as the tracer. Tracer was administered as a 15 s bolus infusion followed by a 90 s static List Mode activity acquisition in 3D mode. Arterial blood activity was measured using a peristaltic roller pump and BGO crystal detector system (Allogg) cross-calibrated with the scanner. All [15O]H2O scans were corrected for detector dead time, tracer decay, and measured tissue photon attenuation, and reconstructed into a 128 × 128 matrix using a 3D transaxial Hann filter with a 4.6 mm cutoff. The field-of-view was 30 cm, resulting in a pixel size of 300 mm/128 = 2.34 mm. Photon attenuation in the tissue was taken into account according to transmission scans obtained using three external circulating 68Ge rod sources. Individual 1.5 T MRIs were acquired for neuroanatomy (Gyroscan Intera CV Nova Dual; Philips Medical Systems).
Parametric images and preprocessing
[15O]H2O images were computed into quantitative parametric rCBF images using arterial blood tracer activity data and tracer kinetic modeling (Kaisti et al., 2002). Images were then realigned, coregistered, and spatially normalized to Montreal Neurological Institute space using the processing routines implemented within the Statistical Parametric Mapping (SPM) software package (SPM-8; Wellcome Department of Cognitive Neurology, University College London, London, England; available at www.fil.ion.ucl.ac.uk/spm) running under MATLAB 2009b (MathWorks). All normalized PET images were smoothed using an isotropic Gaussian kernel of 12 mm full-width at half-maximum.
Images were analyzed for overall patterns of condition-related rCBF changes using partial least-squares (PLS) software (PLS-software; http://www.rotman-baycrest.on.ca/index.php?section=84). PLS is a multivariate analysis technique used to detect optimal covariation between brain voxel-values and the experimental design (Lobaugh et al., 2001; Lin et al., 2003). A mathematical description of the method has been given in detail previously (McIntosh et al., 1996; McIntosh and Lobaugh, 2004).
Analyses were directed at four primary issues. First, to identify brain areas that changed their function with the changing state of consciousness, independent of drug effects, we analyzed the three scans (LOC, ROC, and LOC-2) obtained from the dexmedetomidine group (n = 8, due to one missing and one damaged scan). Second, as some suggest that the anesthesia-induced change in consciousness is related to changing functional connectivity within the frontoparietal network (Imas et al., 2005; Alkire et al., 2008; Boveroux et al., 2010; Boly et al., 2011; Ku et al., 2011), we assessed the changes in parietal functional connectivity between the ROC and the LOC-2 scans of the dexmedetomidine group using seed PLS (McIntosh et al., 2003). These particular scans were the least influenced by changes in drug levels or extemporaneous stimulation effects. Third, to understand the generalizability of the results across agents, the four scans (BL, Sed, LOC, ROC) obtained from the propofol group were analyzed for propofol-induced effects on consciousness (n = 9, due to a missing scan). Fourth, a combined analysis of the propofol and the dexmedetomidine groups was performed to reveal the condition-related (rather than drug-specific) effects on consciousness (n = 17).
The statistical significance of each latent variable (LV) arising from the various PLS analyses was determined using permutation tests (McIntosh et al., 1996). In the present study, 1000 permutations were computed. The reliability of the identified saliences was assessed by a bootstrap estimation of the standard error (SE; 1000 iterations), which provides evidence of the stability of the findings rather than simply demonstrates whether an effect exists (McIntosh and Lobaugh, 2004). Voxels with saliences >3.3 times their SE (corresponding to an approximate p < 0.001 on a two-tailed normal distribution) were considered reliable. As saliences are identified in one single analytic step, correcting for multiple-comparisons is not necessary in PLS. For clarity, only positive saliences are presented in the figures. The SPM-8 glass brain template was used in 3D visualization of the PLS analysis results. However, PLS and not SPM was used to generate these findings.
No adverse events or clinically significant changes were observed in the vital parameters (data not shown). Condition-related drug blood levels are shown in Table 1.
Imaging the ROC during constant level dexmedetomidine
All subjects in the dexmedetomidine group were successfully aroused from unconsciousness into the conscious state with gentle tactile and/or loud verbal stimulation without any alteration in the dosing of the drug. All but one of the subjects subsequently returned to an unconscious state after they were left unstimulated for a few minutes. Multivariate pattern analysis of the PET scans obtained in the three conditions (LOC, ROC, and LOC-2) on dexmedetomidine resulted in three LVs with only LV1 being significant (p < 0.0001) and accounting for 80% of the cross-block covariance between the change in the level of consciousness and the brain activity patterns. The pattern of changes identified in LV1 is shown in Figure 2 and Table 2. The presence of a conscious state was specifically related to activation in a number of brain areas including: the anterior cingulate cortex (ACC) in the medial frontal lobe, the midline thalamus, hypothalamus, the locus ceruleus/parabrachial area in the brainstem, cerebellum, and portions of the lateral orbital frontal and parietal lobes (Fig. 2b–d). The interpretation of this result is aided by the representative voxel intensity plots provided in Figure 2e. These plots show how the identified areas follow the overall pattern of LV1.
Functional connectivity and consciousness
We further investigated how the change in the state of consciousness and the ability to respond affected the functional connectivity of the parietal cortex when the subjects on dexmedetomidine again changed from the conscious, but sedated condition (ROC), to the second unconscious episode (LOC-2). As presented in a correlation plot difference image in Figure 3 and Table 3, the inferior parietal cortex (Brodmann areas 39/40; Fig. 2c,d, marker iii) was functionally connected with the rostral ACC and other frontal regions more in the conscious state compared with the unconscious state.
Arousal from propofol-induced unconsciousness
To understand the generalizability of the findings seen with dexmedetomidine, we analyzed the four sequential PET scans obtained from the subjects who received propofol-TCI in a stepwise increasing manner at the four different levels of consciousness (BL, Sed, LOC, and ROC). PLS analysis of these four conditions yielded four LVs associated with the changing levels of consciousness. The first and the second latent variables (LV1 and LV2) accounted for 68% (LV1, p < 0.0001) and 23% (LV2, p < 0.03) of the cross-block covariance. The third latent variable (LV3) was not significant and the fourth latent variable (LV4) represented the residuals. The relevant results for the comparison with the dexmedetomidine findings are those found with LV2 (Fig. 4). With propofol, the ROC was associated with the activation of anterior arousal networks of the thalamus, brainstem, hypothalamus, and the dorsal ACC (Fig. 4b,c,e). Additional activation was observed in the precuneus/posterior cingulate region (Fig. 4b,c). Compared with the dexmedetomidine results, the emergence from propofol anesthesia showed even less neocortical activation (Fig. 4c,d).
Combined analysis of arousal from unconsciousness
To reduce drug-specific effects on the results and to further dissociate the brain activity changes related to the different states of consciousness from those more likely to be specifically associated with either dexmedetomidine or propofol, we combined the groups for a condition-related PLS analysis. This analysis used the scans from the four states (BL, Sed, LOC, and ROC) and yielded four LVs associated with the changing levels of consciousness. The first and the second latent variables (LV1 and LV2) accounted for 68 and 25% of the cross-block covariance, respectively (p < 0.0001, for both). The third latent variable (LV3) was not significant and the fourth latent variable (LV4) again represented the residuals.
Figure 5 and Table 4 shows the relevant arousal-related LV2 pattern of activity, which was associated with the return of a conscious state following drug-induced unconsciousness. Activation was again seen in the anterior brain arousal networks (i.e., thalamus, hypothalamus, the locus ceruleus/parabrachial area, and the ACC) when unconscious subjects regained their consciousness and became responsive to external stimuli. In this combined analysis, the ROC was not associated with a significant restoration of cortical activation (Fig. 5d).
Combined analysis of drug-related suppressions associated with anesthesia
The other activity pattern (LV1) found in the PLS analysis of combined groups (Fig. 6 and Table 4) revealed suppressed regional brain activity in bilateral frontoparietal regions, the precuneus/posterior cingulate, the ACC in the medial prefrontal cortex, the thalamus, and in the brainstem (Fig. 6b–d). Representative voxels that follow this pattern of activity are shown in Figure 6e. The LV1 patterns for each anesthetic separately displayed similar regional findings (data not shown).
Most neuroimaging studies of anesthetic action assume that loss of functional activity reflects the neural correlates of consciousness (Heinke and Schwarzbauer, 2002; Alkire et al., 2008; Brown et al., 2010). Yet, this approach lacks the ability to distinguish the consciousness-related changes from the epiphenomenal, e.g., drug-related, effects. Here, we used a different approach and investigated what turns back on when consciousness re-emerges following anesthetic-induced unconsciousness. The structures that activated when consciousness resumed were the brainstem, the thalamus, the hypothalamus, and the ACC (Figs. 2, 4, 5). Although only limited frontal and parietal activity was observed during the ROC, an active parietal region demonstrated greater functional connectivity with the ACC (and other frontal regions) during the conscious state (Fig. 3). These findings reveal a functional network that activates with restored consciousness to enable arousal, the subjective awareness of stimuli, and the behavioral expression of the contents of consciousness (Fig. 1).
Arousal-induced activations were mostly localized in deep, phylogenetically old brain structures rather than in the neocortex. These regions reactivated upon returning consciousness in subjects receiving a constant dexmedetomidine infusion (Fig. 2) but also in subjects recovering from propofol-induced unconsciousness (Fig. 4), suggesting that these deep brain structures form a common foundation for a conscious state. These structures activate also upon awakening from natural Stage 2 sleep (Balkin et al., 2002), and their direct stimulation can reverse the unconsciousness of anesthesia (Alkire et al., 2007) or improve behavioral arousal following chronic brain injury (Schiff et al., 2007). On the other hand, impaired consciousness or coma is known to result from paramedian thalamic and midbrain infarcts (Castaigne et al., 1981).
The arousal findings (LV2) associated with propofol (Fig. 4) were consistent with the dexmedetomidine state-related findings (Fig. 2). The separate analyses for both anesthetic agents identified brainstem, thalamus, hypothalamus, and ACC as regions reactivating upon returning consciousness. With propofol, the concentration of the drug was decreasing as subjects regained consciousness, suggesting the regional findings might be attributable to a changing drug level. However, as there was no change in drug dosing with dexmedetomidine and there was strong similarity of findings between dexmedetomidine and propofol, it seems likely that the observed changes were due to an altered state of consciousness and not to any particular interaction with drug dose. The only difference in the regional findings between these drugs was the activation of precuneus/posterior cingulate region during the recovery from propofol anesthesia, which could suggest a propofol-specific effect on this brain region.
Xie et al. (2011) also examined the gain of function associated with the restoration of consciousness during constant dose anesthesia. By using physostigmine, a centrally acting anticholinesterase drug, they were able to increase brain acetylcholine levels and reverse propofol-induced unconsciousness. In subjects who responded, thalamus, lateral parietal cortex, and precuneus reactivated with restored consciousness. Thus, in their study and in the present study, a conscious state was associated with the activation of the thalamus and the parietal cortex. In our study, however, only the subjects anesthetized with propofol showed precuneal activation upon arousal, further suggesting that the activation of precuneus might be a propofol-specific finding. Another difference between these studies is that Xie et al.(2011) did not find activation of the ACC. As consciousness was restored in the dexmedetomidine subjects without any pharmacological manipulation, it seems possible that the physostigmine used by Xie et al. (2011) could have influenced their regional findings.
Perhaps unexpectedly, the return to a conscious state was not accompanied by large changes in neocortical function. This lack of a cortical response was especially apparent with propofol (Fig. 4d) and in the combined analysis of the propofol and dexmedetomidine scans (Fig. 5d). This phenomenon may be more understandable from a clinical perspective. The recovery from anesthesia does not occur all at once, but rather it appears to occur in a bottom-up manner. When emerging from deep anesthesia there will first be signs of autonomic arousal, followed by a slow return of brainstem reflexes, eventually leading to reflexive or uncoordinated somatic movements that occur somewhat before subjects can willfully respond to simple commands. As shown in our results, only minimal cortical activity is necessary at this point. Thus, emergence of a conscious state, the essential foundation of consciousness (Fig. 1), precedes the full recovery of neocortical processing required for rich conscious experiences. We propose that the failure of simple processed EEG monitoring technology to detect patient awareness during anesthesia is at least partly due to the aforementioned pattern of brain arousal (Avidan et al., 2008, 2011; Kaskinoro et al., 2011).
All of these data are in agreement with the experiences obtained from hydranencephalic children, who are devoid of nearly the entire cerebral cortex and yet still display conscious-like behavior (Merker, 2007). Although these children have clear deficits in experiencing the rich contents of consciousness, they undoubtedly are in a conscious state, supporting the idea that the subcortical areas identified in our study play a fundamental role in consciousness. Indeed, numerous studies indicate that consciousness as a process involving both the conscious state (Schiff and Plum, 2000; Lydic and Baghdoyan, 2005) and the contents of consciousness (Baars et al., 2003) likely arises from the interactions between cortical and subcortical mechanisms working through specific network connectivity within the brain (Baars, 1995; Tononi, 2004; Xie et al., 2011).
A number of studies have now examined the connectivity effects associated with changing levels of consciousness (Imas et al., 2005; Massimini et al., 2005; Boveroux et al., 2010; Mhuircheartaigh et al., 2010; Stamatakis et al., 2010; Boly, 2011; Boly et al., 2011; Schrouff et al., 2011). It has been hypothesized that decreased frontal to parietal feedback is a likely correlate for the diminished consciousness associated with sleep (Massimini et al., 2005), brain injury (Boly et al., 2011), and anesthesia (Imas et al., 2005; Alkire et al., 2008; Ku et al., 2011). Our functional connectivity findings with dexmedetomidine (Fig. 3) revealed that the inferior parietal area correlated with frontal activity in the conscious state, but not so in the unconscious state. This demonstrates that functional connectivity exists between frontal and parietal brain areas when the subjects are conscious, but that connectivity fades with LOC. This is not a drug-dose driven change in functional connectivity because the drug concentration remained constant. Our findings are thus consistent with the emerging idea that frontoparietal connectivity plays some role in conscious information processing (Alkire et al., 2008; Ku et al., 2011).
Suppressed frontoparietal and thalamic activity appear to be a common finding in most anesthesia studies (Kaisti et al., 2002; Baars et al., 2003; Bonhomme et al., 2008; Mhuircheartaigh et al., 2010; Stamatakis et al., 2010; Schrouff et al., 2011; Tu et al., 2011). These changes have generally been interpreted directly as neural correlates for LOC. However, since these anesthetic-induced regional deactivations could simply represent the brain areas most affected by the drug used to induce unconsciousness, establishing a causal relationship between these changes and consciousness is difficult. Indeed, pattern analysis across the four states of consciousness revealed that these brain regions do follow the overall pattern of drug level changes (Fig. 6), and should thus be interpreted with caution.
One distinct cortical region activating during the recovery of consciousness was the ACC (Figs. 2⇑⇑–5). This medial prefrontal region has been proposed to play a critical role in consciousness by integrating cognitive-emotional processing with the state of arousal and the intent-to-act (Paus, 2001). Others suggest that the ACC is a key site of self-regulation of behavior (Posner et al., 2007), or might even be involved with the feeling of “free will” (Crick, 1994). The neuroanatomy of the ACC and its connections with motor control regions suggest that this brain structure acts as the neural interface for translating intentions into actions. This idea is in agreement with studies showing that restored activity in the ACC correlates with the level of responsiveness in brain-injured patients (Qin et al., 2010; Zhou et al., 2011). Our inferior parietal finding may have importance because of its functional connection with the ACC during the conscious state and because of its suggested role in movement intention and motor awareness (Desmurget and Sirigu, 2009). In work presented by Desmurget et al. (2009), direct electrical stimulation of the inferior parietal cortex (Brodmann areas 39/40) produced in awake brain surgery patients the sensation of a “will to move.” Desmurget and colleagues (2009) concluded that conscious intention and motor awareness arise from increased parietal activity before movement execution. This idea fits well with our findings.
In summary, we studied the emergence of consciousness, as manifested by the return of its behavioral expression. We have shown that ROC after a period of anesthetic-induced unconsciousness involves activation of the brainstem, thalamus, and the ACC arousal networks as well as restored functional connectivity within the frontoparietal network.
This work was funded by the Academy of Finland (Research Programme on Neuroscience, NEURO, project no. 8111818), Turku University Hospital EVO-Grant no. 13323, Seinäjoki Central Hospital EVO-Grant no. 1133, The Finnish Medical Foundation, and The Paulo Foundation. Presented in part at the American Society of Anesthesiologist's annual meeting, Best abstracts of the meeting session, October 16–20, 2010, San Diego, California.
The authors declare no competing financial interests.
- Correspondence should be addressed to Harry Scheinin, Turku PET Centre, University of Turku, and Turku University Hospital, P.O. Box 52, FI-20521 Turku, Finland.
This article is freely available online through the J Neurosci Open Choice option.