European Journal of Pain Supplements 5 (2011) 323–327

Contents lists available at SciVerse ScienceDirect

European Journal of Pain Supplements

journal homepage: www.EuropeanJournalPain.com

How neuroimaging can help us to visualise and quantify pain?

Karolina WartolowskaNuffield Division of Anaesthetics, Nuffield Department of Neurological Sciences, University of Oxford, Oxford, UK

a r t i c l e i n f o

Keywords:Neuroimaging

PainFunctional connectivityFMRIASL

1754-3207/$36.00 � 2011 European Federation of Intdoi:10.1016/j.eujps.2011.08.012

E-mail address: karolina.wartolowska@nda.ox.ac.u

a b s t r a c t

Pain is a complex and multidimensional experience, which is subjective for an individual and modulatedby physiological and psychological factors. Therefore, it is difficult to quantify pain and there are noobjective pain measures available at the moment. Neuroimaging provides an objective measure ofchanges in brain activity related to pain perception. In this review, we demonstrate that pain-relatedbrain activation is complex and can be best studied as a dynamic network of interconnected regions.Finally, we use the placebo and nocebo effects to discuss the factors involved in modulation of painexperience.

� 2011 European Federation of International Association for the Study of Pain Chapters. Published byElsevier Ltd. All rights reserved.

1. Neuroimaging techniques

Neuroimaging is a valuable tool for visualising pain and it hasimmensely increased our knowledge of pain processing.

There are several neuroimaging methods typically used toinvestigate the neural mechanisms involved in pain processing,including functional magnetic resonance imaging (FMRI), positronemission tomography (PET), magnetoencephalography (MEG) andelectroencephalography (EEG). There are also other techniques thatmeasure structural or metabolic changes in the brain such as diffu-sion tensor imaging, morphometric imaging and spectroscopy.

The most popular technique used for studying pain-relatedbrain activation is FMRI. FMRI provides information about neuralactivity within the whole brain with anatomical accuracy and iscompletely non-invasive. This technique does not require exoge-nous contrast agents or radioligands as it uses magnetic propertiesof blood to create image contrast. Although this is not a directmeasure of brain activation, it has been demonstrated that FMRIsignal correlates well with neuronal activity (Logothetis, 2003).There are two FMRI methods commonly used: blood oxygenationlevel dependent (BOLD) and arterial spin labelling (ASL) (Raichleand Mintun, 2006; Tracey and Mantyh, 2007).

BOLD FMRI measures local changes in blood oxygen contentthat normally take place in an active part of the brain as a resultof an increased delivery of oxygen in response to an increase inmetabolism. This creates the BOLD contrast. This method is notan absolute measure as it measures signal change relative to base-line. BOLD FMRI is typically used for studying brain activity inresponse to evoked brief stimuli and cannot be reliably used tostudy ongoing pain because of the need for a baseline and limits

ernational Association for the Stud

k

to stimulus duration. In a typical experiment, brief (2–3 s) painfulstimuli such as pin-prick or heat are delivered alternating withperiods of rest (about 30–60 s long) when no pain is present. Stim-uli are usually repeated 10–15 times to increase a signal to noiseratio. Then, statistical methods are used to analyse how well thesignal in different brain regions matches the timing of the painstimuli. The closer the match, higher the value on the statisticalmap representing the activation. Therefore, in BOLD FMRI a magni-tude of activation refers to a spatial extent of the significant signalchanges and our confidence, i.e., a statistical significance, thatchanges in those regions result from the stimulation.

ASL uses magnetically labelled blood as a contrast. Changes inASL FMRI signal correspond to changes in brain perfusion. Depend-ing on the technique, it allows us to measure either relative or abso-lute values of blood perfusion in ml of blood per 100 g of braintissue per minute. In contrast to BOLD FMRI, ASL FMRI perfusion re-sults represent not a statistical parameter but an actual physiolog-ical variable. Moreover, it does not require a change in stimulus andcan be used for imaging activation lasting minutes. Therefore, it issuitable for studying the neural mechanisms of tonic evoked painor clinical ongoing pain (Buxton, 2005; Tracey and Johns, 2010).

2. Brain regions involved in pain perception

The pathways and brain regions involved in pain processinghave been relatively well-characterised using electrophysiologicalmethods, histological tracers or lesion studies (Guilbaud et al.,2005). Neuroimaging provides an insight into the neurophysiologi-cal basis of pain not previously possible with other methods.

Neuroimaging established the role of the brain in pain by dem-onstrating extensively that the central nervous system is crucial forcreating pain perception. It is important to remember that there is

y of Pain Chapters. Published by Elsevier Ltd. All rights reserved.

324 K. Wartolowska / European Journal of Pain Supplements 5 (2011) 323–327

no single brain region that encodes ‘‘pain’’ and there are severalregions that show somatotopy for pain (Baumgartner et al., 2010;Bingel et al., 2004). What becomes apparent is that there is no ‘‘pri-mary pain cortex’’ for pain, an equivalent of the ‘‘primary sensorycortex’’ for touch (Tracey, 2011). There is no single region that,when dissected, will cause the pain to stop, although we know thatlesion to the insular cortex may alter the pain sensation (Green-span et al., 1999). There is probably not even a ‘‘primary nocicep-tive cortex’’. There are several regions however, including theposterior part of the insular cortex, the operculum or the secondarysomatosensory cortex, that are possible candidates (Baumgartneret al., 2010; Tracey, 2011). The posterior insular cortex is particu-larly interesting as it encodes stimulus intensity across differentmodalities of pain and is only activated when an actual noxiousinput is present (Tracey, 2011).

As pain is a complex and multidimensional experience, it is rep-resented by several regions within the brain (Apkarian et al., 2005;Tracey and Mantyh, 2007). Painful stimulation typically results inbilateral activation in the primary and secondary somatosensorycortices, the thalamus, the insular cortex and the anterior cingulatecortex. Activation in these regions is thought to be related to thesensory-discriminative aspects of pain processing, while the pre-frontal and parietal cortices appear to participate in the cognitiveaspect of pain and paradigms involving threat or anxiety usuallyevoke responses in the hippocampus and amygdala (Peyronet al., 2000; Tracey and Mantyh, 2007). It is important to empha-sise the fact that there is no fixed set of regions responding to pain,there is no definitive ‘‘pain matrix’’ that is always present duringpain perception. In addition, the pain-related regions are also notunique to pain experience (Tracey and Johns, 2010). Brain struc-tures that are involved in pain processing are also engaged in pro-cessing and modulation of other sensory modalities as well asother emotional or cognitive states. As pain is multidimensional,its processing involves regions encoding not only the sensory butalso the emotional, cognitive and motivational aspects of pain.

3. Brain regions encoding intensity of pain

Quantifying pain poses a challenge, as there is no objectivemeasure of pain. Behavioural measures that are commonly usedfor characterising pain have several limitations as the patient isbeing asked to give a single number on a uni-dimensional scaleto describe a complex experience of pain. In addition, the reportis usually verbal which may be another limiting factor.

It is important to keep in mind that report bias is not just anartefact but an effect of pain modulation by internal, i.e., cognitive,emotional, genetic, and external factors such as stimulus proper-ties (Chizh et al., 2009; Tracey and Mantyh, 2007).

The perceived and reported pain is often not linearly related tothe strength of the external input. There is a complex relationshipbetween the nociceptive input and the pain experience resultingfrom it. In addition to that, nociception is not necessary to evokepain perception, i.e., patients may feel pain without any externalinput for example in post stroke pain.

Neuroimaging offers an objective way of investigating pain byobserving changes in brain activity related to pain. Brain imagingstudies have demonstrated that the magnitude of pain-relatedactivation is related to the intensity of the input and perceived pain(Coghill et al., 1999; Tracey and Mantyh, 2007).

As mentioned, the operculo-insular region is crucial for painperception (Coghill et al., 1999); however, pain representation inthat region is complex and there are several somatotopic represen-tations of pain in that region (Baumgartner et al., 2010). Thesensory-discriminative aspect of the painful stimulation is alsoencoded in the primary and secondary somatosensory cortices,

and the anterior cingulate cortex (Coghill et al., 1999) as well asin the thalamus, putamen and cerebellum (Bingel et al., 2002).The primary somatosensory cortex and the anterior cingulate cor-tex activation reflect specifically the intensity of pain experiencerather than the input strength (Lee et al., 2008). On the other hand,information about stimulus intensity is encoded in the operculum,the insular cortex as well as the secondary somatosensory cortex;therefore, it may not be specific to pain (Baumgartner et al., 2010).In fact, stimulus intensity processing engages many cortical andsubcortical regions, not only those encoding the sensory aspectof pain but also regions involved in attention, emotions and motorcontrol (Coghill et al., 1999). This demonstrates that intensity isalso relevant for other processes. The fact that intensity encodingis preserved across different brain regions supports the hypothesisof parallel pain processing (Coghill et al., 1999).

4. Pain processing in terms of functional connectivity

Perception is a dynamic process that is constantly modulated ina top-down fashion by attention and expectations. Top-down influ-ence means that even the earliest stages of cortical sensoryprocessing are influenced by the higher cognitive processes (Gil-bert and Sigman, 2007). Therefore, perception is better reflectedby networks, than by brain regions in isolation. In order to under-stand the mechanisms involved in pain perception we have tostudy not just these regions, but connections between thoseregions and with other networks (Ploner et al., 2011).

Neuroimaging studies have demonstrated that the functionalconnectivity between the pain-related regions and other brainnetworks determines the subsequent pain experience (Ploneret al., 2010; Tracey and Johns, 2010).

Functional connectivity tells us about changes in the strength ofnetwork connections between brain structures (Cauda et al., 2009).Those dynamic interactions between the cortical and subcorticalstructures are crucial for pain perception and modulation (Bingelet al., 2006; Valet et al., 2004). Moreover, the connectivity preced-ing the stimulus determines pain perception (Ploner et al., 2010).Ploner et al. (2010) demonstrated that the connectivity betweenthe insular cortex and the brainstem predicts whether a potentiallynoxious stimulus will be perceived as painful. This means thatexpectations bias perception (Keltner et al., 2006). In other words,the neural processing of the stimulus depends on the prior knowl-edge of the stimulus and the adequate adjustment preceding it.Neuroimaging studies have demonstrated that expectation of painis associated with activation of pain-related regions such as theprimary somatosensory cortex, the anterior cingulate cortex andthe insular cortex (N.B. this means that there is activation withinthe pain-related regions in the absence of actual noxious stimulus)(Fairhurst et al., 2007; Ploghaus et al., 1999). The anterior insularcortex plays a key role in adjustment during the anticipation peri-od and is involved in subsequent changes in pain processing(Ploner et al., 2010). It integrates information about the salienceand potential threat of the upcoming stimulus. Moreover, thepre-stimulus activation in the anterior insular cortex results in apotentially painful stimulus (close to pain threshold) beingperceived as actually painful (Wiech et al., 2010). The activationin the medial prefrontal cortex and the anterior cingulate cortexduring the anticipatory period may inhibit the nociceptive trans-mission through cognitive appraisal, i.e., evaluation of meaningof the emotional stimuli (Kalisch et al., 2006; Lorenz et al., 2003).

5. Pain modulation by attention and mood

Neuroimaging has greatly advanced our knowledge of mecha-nisms involved in this modulation as it allowed us to use flexible

K. Wartolowska / European Journal of Pain Supplements 5 (2011) 323–327 325

paradigms to study different factors influencing pain experience(Tracey and Mantyh, 2007; Wiech et al., 2008).

Several neuroimaging studies have demonstrated that distrac-tion changes the sensory aspect of pain and reduces pain ratings.These studies demonstrated that distraction is associated with adecrease in activity in the pain-related regions such as the anteriorcingulate cortex, the thalamus and the insular cortex as well aswith an increase of activity in the pain-modulatory regions, suchas the anterior cingulate and the prefrontal cortex (Bantick et al.,2002). Distraction was also associated with changes in networksand strength of the functional coupling between the anteriorcingulate cortex and the prefrontal cortex as well as the periaqu-eductal grey (PAG) and posterior thalamus (Valet et al., 2004). Anincrease in activation in the periaqueductal grey predicted changesin reported pain intensity (Tracey et al., 2002). This suggests that,during distraction from pain, the anterior cingulate cortex andthe prefrontal cortex exert a top-down influence on the thalamusand brainstem (Valet et al., 2004). The effect of distraction is mostlikely mediated via the descending pain modulatory system byinhibiting the nociceptive processing at the level of the dorsal horn(Tracey and Mantyh, 2007).

Depressed mood also increased the perceived pain intensity.Berna and colleagues showed that experimentally induced sadmood may alter pain perception in healthy people via the emo-tional regulatory networks (Berna et al., 2010). Sad mood resultedin stronger brain activation in response to pain in the regions pro-cessing pain and negative emotions, such as the anterior cingulatecortex and the hippocampus. There was also an increase in theactivation in the prefrontal cortex related to appraisal (Bernaet al., 2010).

In a recent experiment Ploner et al. (2011) demonstrated thatattention and emotions both modulate pain but through differentmechanisms and they are related to different patterns of functionalconnectivity. Both attending to the stimulus and negative moodaugmented pain and were associated with higher pain ratings.The magnitude of pain ratings correlated with activation of theanterior insular cortex. Interestingly, the attentional modulationaltered the functional connectivity between the anterior insularcortex and frontoparietal regions whereas the emotional modula-tion changed the connectivity between the anterior insular cortexand limbic structures such as parahippocampal gyrus and amyg-dala. This suggests that the contextual modulation of pain dependson flexible functional interactions between the anterior insularcortex and the frontoparietal or the mediotemporal. Moreover,the modulation converged on a common structure, i.e., the anteriorinsular cortex. This suggests that the anterior insular cortex isimportant for integration of the sensory input and emotional andcognitive factors for the higher order representation of a state. Thisis in line with a study by Wiech and colleagues who demonstratedthat the anterior insular cortex encodes stimulus salience therebyaltering pain perception (Wiech et al., 2010). In a recent reviewWiech et al. (2008) suggested that the descending modulatory sys-tem might interact or overlap with a more general system forattentional control that is not specific for pain.

6. Placebo and nocebo modulation

Neuroimaging studies on the placebo and nocebo effects are anexcellent example demonstrating the complexity of pain process-ing and the mechanisms involved in modulating pain.

It is accepted that placebo is not a response/report bias (Hrobj-artsson and Gotzsche, 2001). It is a result of interactions betweenthe psychological and physiological effects of therapeutic proce-dures even in the absence of pharmacologically active analgesicsubstances. The placebo/nocebo effect has both unconscious

components (conditioning) as well as conscious components (cog-nition, expectation, belief) (Bingel et al., 2011).

Several studies (Wager et al., 2004) used neuroimaging to inves-tigate the neural mechanisms of placebo/nocebo effect. Placebomanipulation was associated with a decrease of activation inpain-related regions (Bingel et al., 2006, 2011), whereas nocebo ef-fect was related to an increase of pain-related activity (Bingel et al.,2011). The magnitude of the activation change correlated with thechange in the reported pain intensity which confirmed thatplacebo is a real effect not just a report bias (Bingel et al., 2011).

Neuroimaging studies demonstrated that placebo response iscomplex and may act via several mechanisms (Wiech et al.,2008) including: expectations about pain (Wager et al., 2004), cog-nitive appraisal (Petrovic et al., 2002), attentional control or thedescending modulatory system including the opioid-related mod-ulation and/or inhibition of afferent transmission (Wiech et al.,2008).

Placebo alters pain experience through modulation of expecta-tion as has been demonstrated by Wager et al. (2004) who showedan increased activation in the prefrontal cortex during anticipationof pain. The magnitude of the placebo effect correlated with theactivation during the anticipation period in the orbitofrontalcortex, the dorsolateral prefrontal cortex and the rostral anteriorcingulate cortex (Wager et al., 2004). This suggested that therostral anterior cingulate cortex and the prefrontal cortex were in-volved in cognitive (expectation) control of pain processing relatedto the placebo effect and pain relief (Wager et al., 2004). The func-tional connectivity between the rostral anterior cingulate cortexand the prefrontal cortex is directly associated with a magnitudeof pain relief related to the placebo effect (Petrovic et al., 2005;Wager et al., 2004). They also showed that the strongest placeboresponders show the strongest coupling between the rostral ante-rior cingulate cortex and the prefrontal cortex (Petrovic et al.,2005). The prefrontal cortex activity was associated with a reduc-tion of pain (Lorenz et al., 2003; Seifert et al., 2009) probably viathe top-down modulation of pain processing (Petrovic et al.,2010) related to cognitive appraisal of pain (Wiech et al., 2008).In addition, the orbitofrontal cortex is involved in expectationsand attributing a specific value to external stimuli (Petrovic et al.,2005). From neuroimaging studies (Petrovic et al., 2005) we knowthat both emotional placebo and placebo analgesia act through therostral anterior cingulate cortex and the orbitofrontal cortex. Ven-tro-orbital region was involved in top-down cognitive modulatoryprocesses that might not be pain specific (Petrovic et al., 2002).This reflects a complex modulatory process that is not specific toplacebo analgesia.

During the anticipation period, the activation in the prefrontalcortex and the anterior cingulate cortex correlated with the activa-tion in the brainstem supporting the idea that prefrontal mecha-nisms are involved in top-down modulation of pain and the painmodulatory system within the brainstem for placebo and opioidanalgesia (Petrovic et al., 2002; Wager et al., 2004). The descendingmodulatory system is crucial for the cognitive and psychologicalcontrol of pain (Wiech et al., 2008). Wiech suggested that thedescending modulatory pathway may be the final common path-way for analgesia or hyperalgesia (Wiech et al., 2008).

Bingel et al. (2011) used neuroimaging to investigate the neuralmechanisms by which expectations, both positive and negative,interact with the physiological effects of an active drug (remifenta-nil). Behaviourally, placebo augmented the effect of the active drugwhereas nocebo cancelled the analgesic effect. This was associatedwith a decrease and an increase of pain-related activation, respec-tively. Changes in pain rating (despite constant stimulus strength)were predicted by changes of activation in the rostral anteriorcingulate cortex, i.e., increased activity during placebo and deacti-vated during the nocebo condition. As it has been mentioned

326 K. Wartolowska / European Journal of Pain Supplements 5 (2011) 323–327

earlier, the rostral anterior cingulate cortex plays an important rolein placebo and opioid analgesia (Petrovic et al., 2005). In contrast,the hippocampus was active only during nocebo and was associ-ated with increased anxiety ratings suggesting that it may berelated to an impairment of analgesia due to negative expectations.This study demonstrated that expectation about the effect of thedrug engaged mechanisms responsible for descending modulationof pain.

Several studies demonstrated that placebo analgesia engagesthe endogenous pain modulatory opioid network and leads todecreased afferent input to the brain (Petrovic et al., 2002; Wageret al., 2004). However, this is not the only mechanism and the pla-cebo effect is complex and dynamic (Petrovic et al., 2010).

Neuroimaging was used to investigate whether placebo actsearly through expectation, or late through opioid release, inresponse to pain or cognitive reappraisal of pain. Wager and col-leagues examined temporal changes related to the placebo effectand observed an early change in the rostral anterior cingulatecortex suggesting that it is related to expectations. Other pla-cebo-related changes were observed late suggesting that theymay be associated with cognitive appraisal or opioid release inresponse to pain (Wager et al., 2004).

It has also been demonstrated that placebo analgesia shows sig-nificant differences from opioid analgesia despite involvement ofopioidergic system in both. In the opioid condition, expectationfor analgesia was met and was congruent with the level of nocicep-tive processing. In the placebo condition, the expectation wasincongruent with the processing level of the nociceptive input(Petrovic et al., 2010).

In this study, the opioid rich rostral anterior cingulate cortexwas more active during the opioid condition than the placebo con-dition. In contrast, the orbitofrontal cortex and the prefrontal cor-tex were more activated during the placebo condition probablyreflecting the cognitive regulation of pain. Also the functional con-nectivity between rostral anterior cingulate cortex and prefrontalcortex was stronger in the placebo condition than during the opi-oid condition and the degree of connectivity was correlated withthe degree of the placebo response. They interpreted this in termsof expectations and signal error. This supports the hypothesis thatperception is a result of external incoming signals as well as theinternal expectations (Petrovic et al., 2010).

7. Visualising ongoing pain

The majority of functional imaging studies used BOLD FMRI andbrief noxious stimuli; therefore, they were limited to studyingacute pain. Acute pain is not necessarily the same as tonic pain.Understanding the mechanisms involved in tonic pain is importantfor studies of clinical ongoing pain.

ASL offers a tool to examine clinical pain and the dynamics ofpain processing during tonic or ongoing pain and subsequently im-prove understanding of pain processing and neural mechanismsinvolved in clinical pain.

Howard et al. (2011) used ASL to study actual clinical pain, i.e.,post-operative pain associated with tissue damage. They demon-strated that ongoing pain was associated with an increase of acti-vation bilaterally in the pain-related regions and the magnitude ofactivation correlated with pain ratings. The pain-related activationwas stable during ongoing pain and was reproducible betweensessions.

Owen et al. (2010) used ASL to study a tonic muscular painexperimental model (intramuscular injection of hypertonic saline)in healthy volunteers. They investigated the temporal changes inthe pattern of activation, i.e., the dynamics of pain evolution asthe pain evolved from the first sharp acute pain to tonic pain.

The transition from acute to tonic pain was associated with a de-crease in the number of regions activated as well as a decrease inthe spatial extent of the activation mostly in the brain regionsinvolved in the processing of affective and cognitive dimensionsof pain. Different rates of activity changed over time suggestingthat different aspects of pain might evolve differently during ongo-ing pain. The insular cortex was the only region that showed a sus-tained activation during the tonic pain and activation in that regioncorrelated strongly with pain intensity ratings. This study showedthat ASL can be used to study ongoing pain processing and provideadditional information about the specific mechanisms involved inthis process.

8. Future directions

The big question for neuroimaging in pain research is whetherthese methods can be used as a read-out of pain experience inde-pendent of behavioural ratings. Neuroimaging has the potential tobecome an objective measure of pain and replace subjective report.Using neuroimaging to characterise pain without a need to rely onsubjective pain ratings or descriptions of pain would be also ofgreat benefit for measuring treatment efficacy.

Neuroimaging studies have shown that every individual has aunique pain signature, as a result of emotional and cognitive state,pain memories and pain beliefs, genetic profile, hormones andother factors that affect pain processing. Therefore, neuroimagingcan potentially be used to phenotype pain (Tracey, 2011).

9. Conclusions

Neuroimaging changed the field of pain research. It made it pos-sible to visualise pain processing and modulation in the brain in anon-invasive and non-interfering way. Therefore, it greatly in-creased our knowledge of pain processing at the level above thespinal cord.

Brain imaging gives an insight into the complexity of pain. Painexperience is an outcome of the external input as well as psycho-logical, physiological and cognitive factors, which make pain verysubjective. Therefore the ‘‘cerebral pain signature’’ is unique for aparticular person at that particular moment.

Neuroimaging also provided complementary information aboutpain such as the effects of psychological, cognitive or pharmacolog-ical manipulations or the temporal dynamics of pain processing.Finally, brain imaging studies demonstrated that pain is a complexperception reflected by dynamic and flexible network interactions.

10. Conflict of interest statement

None declared.

References

Apkarian AV, Bushnell MC, Treede RD, Zubieta JK. Human brain mechanisms of painperception and regulation in health and disease. Eur J Pain 2005;9:463–84.

Bantick SJ, Wise RG, Ploghaus A, Clare S, Smith SM, Tracey I. Imaging how attentionmodulates pain in humans using functional MRI. Brain 2002;125:310–9.

Baumgartner U, Iannetti GD, Zambreanu L, Stoeter P, Treede R-D, Tracey I. MultipleSomatotopic Representations of Heat and Mechanical Pain in the Operculo-Insular Cortex: A High-Resolution fMRI Study. J Neurophysiol 2010;104:2863–72.

Berna C, Leknes S, Holmes EA, Edwards RR, Goodwin GM, Tracey I. Induction ofdepressed mood disrupts emotion regulation neurocircuitary and enhancespain unpleasantness. Biol Psychiatry 2010;67:1083–90.

Bingel U, Gaelscher J, Weiller C, Bchel C. Somatotopic representation of nociceptiveinformation in the putamen: an event-related fMRI study. Cereb Cortex2004;14:1340–5.

Bingel U, Lorenz J, Schoell E, Weiller C, Buchel C. Mechanisms of placebo analgesia:rACC recruitment of a subcortical antinociceptive network. Pain 2006;120:8–15.

K. Wartolowska / European Journal of Pain Supplements 5 (2011) 323–327 327

Bingel U, Quante M, Knab M, Bromm R, Weiller. C, Buechel C. Subcortical structuresinvolved in pain processing: evidence from single-trial fMRI. Pain2002;99:313–21.

Bingel, U., Wanigasekera, V., Wiech, K., Ni Mhuircheartaigh, R., Lee, M.C., Ploner, M.,and Tracey, I. (2011). The effect of treatment expectation on drug efficacy:imaging the analgesic benefit of the opioid remifentanil. Sci Transl Med 3,70ra14.

Buxton RB. Quantifying CBF with arterial spin labeling. J Magn Reson Imag2005;22:723–6.

Cauda F, Sacco K, Duca S, Cocito D, D’Agata F, Geminiani GC, Canavero S. Alteredresting state in diabetic neuropathic pain. PLoS ONE 2009;4:e4542.

Chizh BA, Priestley T, Rowbotham M, Schaffler K. Predicting therapeutic efficacy –experimental pain in human subjects. Brain Res Rev 2009;60:243–54.

Coghill RC, Sang CN, Maisog JM, Iadarola MJ. Pain intensity processing within thehuman brain: a bilateral, distributed mechanism. J Neurophysiol1999;82:1934–43.

Fairhurst M, Wiech K, Dunckley P, Tracey I. Anticipatory brainstem activity predictsneural processing of pain in humans. Pain 2007;128:101–10.

Gilbert CD, Sigman M. Brain states: top-down influences in sensory processing.Neuron 2007;54:677–96.

Greenspan JD, Lee RR, Lenz FA. Pain sensitivity alterations as a function of lesionlocation in the parasylvian cortex. Pain 1999;81:273–82.

Guilbaud G, Bernard JF, Besson JM. Brain areas involved in nociception and pain. In:Koltzenburg S, McMahon M, editors. Wall and Melzack’s textbook ofpain. Churchill Livingstone; 2005.

Howard, M., Krause, K., Khawaja, N., Massat, N., Zelaya, F., Schumann, G., Huggins, J.,Vennart, W., Williams, S., TF, R. Beyond patient reported pain: perfusionmagnetic resonance imaging demonstrates reproducible cerebralrepresentation of ongoing post-surgical pain. PLoS One 2011;6.

Hrobjartsson A, Gotzsche PC. Is the placebo powerless? New Engl J Med2001;344:1594–602.

Kalisch RaW, Critchley HD, Dolan RJ. Levels of appraisal: a medial prefrontal role inhigh-level appraisal of emotional material. Neuroimage 2006;30:1458–66.

Keltner JR, Furst A, Fan C, Redfern R, Inglis B, Fields HL. Isolating the modulatoryeffect of expectation on pain transmission: a functional magnetic resonanceimaging study. J Neurosci 2006;26:4437–43.

Lee MC, Zambreanu L, Menon DK, Tracey I. Identifying brain activity specificallyrelated to the maintenance and perceptual consequence of central sensitizationin humans. J Neurosci 2008;28:11642–9.

Logothetis NK. The underpinnings of the BOLD functional magnetic resonanceimaging signal. J Neurosci 2003;23:3963–71.

Lorenz J, Minoshima S, Casey KL. Keeping pain out of mind: the role of thedorsolateral prefrontal cortex in pain modulation. Brain 2003;126:1079–91.

Owen DG, Clarke CF, Ganapathy S, Prato FS, St. Lawrence KS. Using perfusion MRI tomeasure the dynamic changes in neural activation associated with tonicmuscular pain. Pain 2010;148:375–86.

Petrovic P, Dietrich T, Fransson P, Andersson J, Carlsson K, Ingvar M. Placebo inemotional processing-induced expectations of anxiety relief activate ageneralized modulatory network. Neuron 2005;46:957–69.

Petrovic P, Kalso E, Petersson KM, Andersson J, Fransson P, Ingvar M. A prefrontalnon-opioid mechanism in placebo analgesia. Pain 2010;150:59–65.

Petrovic P, Kalso E, Petersson KM, Ingvar M. Placebo and opioid analgesia – imaginga shared neuronal network. Science 2002;295:1737–40.

Peyron R, Laurent B, Garcia-Larrea L. Functional imaging of brain responses to pain.A review, meta-analysis. Neurophysiol Clin 2000;30:263–88.

Ploghaus A, Tracey I, Gati JS, Clare S, Menon RS, Matthews PM, Rawlins JN.Dissociating pain from its anticipation in the human brain. Science1999;284:1979–81.

Ploner M, Lee MC, Wiech K, Bingel U, Tracey I. Prestimulus functional connectivitydetermines pain perception in humans. Proc Natl Acad Sci 2010;107:355–60.

Ploner M, Lee MC, Wiech K, Bingel U, Tracey I. Flexible cerebral connectivitypatterns subserve contextual modulations of pain. Cereb Cortex2011;21:719–26.

Raichle ME, Mintun MA. Brain work and brain imaging. Annu Rev Neurosci2006;29:449–76.

Seifert F, Bschorer K, De Col R, Filitz J, Peltz E, Koppert W, Maihoefner C. Medialprefrontal cortex activity is predictive for hyperalgesia and pharmacologicalantihyperalgesia. J Neurosci 2009;29:6167–75.

Tracey I. Can neuroimaging studies identify pain endophenotypes in humans? NatRev Neurol 2011;7:173–81.

Tracey I, Johns E. The pain matrix: reloaded or reborn as we image tonic pain usingarterial spin labelling. Pain 2010;148:359–60.

Tracey I, Mantyh PW. The cerebral signature for pain perception and its modulation.Neuron 2007;55:377–91.

Tracey I, Ploghaus A, Gati JS, Clare S, Smith S, Menon RS, Matthews PM. Imagingattentional modulation of pain in the periaqueductal gray in humans. J Neurosci2002;22:2748–52.

Valet M, Sprenger T, Boecker H, Willoch F, Rummeny E, Conrad B, Erhard P, Tolle TR.Distraction modulates connectivity of the cingulo-frontal cortex and themidbrain during pain – an fMRI analysis. Pain 2004;109:399–408.

Wager TD, Rilling JK, Smith EE, Sokolik A, Casey KL, Davidson RJ, Kosslyn SM, RoseRM, Cohen JD. Placebo-induced changes in fMRI in the anticipation andexperience of pain. Science 2004;303:1162–7.

Wiech K, Lin C-s, Brodersen KH, Bingel U, Ploner M, Tracey I. Anterior insulaintegrates information about salience into perceptual decisions about pain. JNeurosci 2010;30:16324–31.

Wiech K, Ploner M, Tracey I. Neurocognitive aspects of pain perception. Trends CognSci 2008;12:306–13.