fMRI and Acupuncture: Overview
Since the late 1990s, functional MRI (fMRI) has become one of the principal tools for investigating how acupuncture affects the human brain. More than 1,500 studies published in indexed journals — including NeuroImage, Brain, PAIN, and Cerebral Córtex — document reproducible patterns of brain activation and deactivation during and after stimulation of acupoints.
These studies provided the first objective and biologically plausible evidence that acupuncture has measurable effects on the central nervous system, distinct from simple cutaneous stimulation, expectation of improvement, or placebo effect. Today, it is consensus that acupuncture modulates specific brain networks — particularly those related to pain processing, affective regulation, and the autonomic system.
This article synthesizes the most solid neuroimaging evidence, translating technical terms into language accessible to clinicians and patients who want to understand acupuncture's central mechanisms.
More than 1,500 studies
Acupuncture has one of the largest neuroimaging research bodies among non-pharmacological interventions — comparable to meditation and greater than many physical therapy techniques.
Reproducible brain networks
Activation/deactivation patterns are reproducible across labs and populations: insula, anterior cingulate córtex (ACC), periaqueductal gray, hypothalamus, and limbic system.
Distinguishes acupuncture from placebo
Verum and sham stimulation show statistically distinct activation patterns — though the topic remains actively debated in the literature.
How fMRI Works
BOLD fMRI (blood-oxygen-level dependent) detects changes in regional cerebral blood flow. When neurons are activated, they consume more oxygen; seconds later, reactive vasodilation brings more oxygenated blood to the region. The magnetic difference between oxygenated and deoxygenated hemoglobin generates the BOLD signal, translated into activation maps.
In acupuncture studies, two designs predominate: block-design fMRI, in which the needle is inserted and manually stimulated in cycles alternating with rest, capturing activation in real time; and resting-state fMRI, which measures functional connectivity between brain regions before and after a complete acupuncture session, capturing sustained effects.
Acupuncture is particularly suited to fMRI studies because it can be administered with the patient inside the scanner, and the sensation of Deqi — heaviness, distension, tingling characteristic of insertion at an active point — is quantifiable and correlates with the magnitude of the effects observed.
Main Findings
Meta-analyses of fMRI studies (Huang et al., PLoS One, 2012; Chae et al., J Pain, 2013) and more recent reviews consistently map the central effects of acupuncture across three domains:
Activation of analgesic regions
Periaqueductal gray (PAG), raphe nucleus, hypothalamus, insula. Together they form the endogenous descending pain modulation system — the same network activated by exogenous opioids such as morphine.
Deactivation of the limbic system
Amygdala, hippocampus, and subgenual anterior cingulate córtex show reduced BOLD signal during sustained acupuncture — explaining the anxiolytic and antidepressant effects observed clinically.
Modulation of the pain matrix
Connectivity reorganizes between the primary somatosensory córtex (S1), anterior cingulate córtex (ACC), insula, and thalamus — the regions that encode, integrate, and assign affective meaning to pain.
Measurable autonomic effect
Increased vagal activity (right insular córtex, nucleus of the solitary tract) and reduced sympathetic activity — the neural basis of effects on heart rate, gastrointestinal motility, and anxiety.
Brain Networks Modulated
Resting-state functional connectivity studies show that acupuncture not only activates isolated regions but reorganizes the interaction among large-scale neural networks:
Default Mode Network (DMN)
Network active at rest, linked to introspection, rumination, and self-perception. Hyperconnected in depression and chronic pain. Acupuncture normalizes its connectivity, with reduction of pathological "self-focus".
Salience Network
Anterior insula + anterior cingulate córtex. Detects relevant stimuli and mediates transitions between networks. Centrally modulated by acupuncture — explaining pain-perception reorganization.
Sensorimotor Network
Reorganized in chronic pain conditions (fibromyalgia, low back pain). Acupuncture restores connectivity patterns close to those of healthy controls.
Dorsal Executive Network
Dorsolateral prefrontal córtex. Involved in attention and inhibitory control. Connectivity increases after acupuncture — a potential basis for effects on concentration and emotional regulation.
Real vs Sham — the Debate
A central research question is whether the brain effects of verum (real) acupuncture differ from those of sham acupuncture (simulated placebo: needles that do not penetrate the skin, needling at non-acupoint locations, or superficial insertion without manipulation).
The most sophisticated reading of the literature (Lund & Lundeberg, Acupunct Med, 2006 onward; reviews in Neuroscience and PAIN) recognizes that:
Sham is not a neutral placebo
Inserting needles anywhere on the body generates real sensory stimulation, activates A-delta and C fibers, locally releases mediators, and produces some central effect — which dilutes the contrast with verum acupuncture.
Different activation patterns
Acupuncture at classical points with Deqi activates the PAG, hypothalamus, and posterior insula significantly more than sham. Sham preferentially activates primary somatosensory regions and expectation circuits.
Point specificity
Within-patient studies show distinct brain patterns for different acupoints — indicating somatotopic specificity that goes beyond non-specific sensory stimulation.
Clinical vs neural effects
In conditions like chronic pain, the clinical difference between verum and sham is modest, but the neural difference is more consistent — suggesting both work through partially distinct mechanisms.
Other Techniques: PET, EEG, NIRS
Neuroimaging of acupuncture is not limited to fMRI:
PET (positron emission tomography)
Visualizes the release of specific neurotransmitters. Studies with [11C]-carfentanil have demonstrated endogenous opioid release after acupuncture — confirming that the central opioid system is directly modulated.
EEG (electroencephalography)
Captures millisecond-scale changes in cortical activity. Studies show increased alpha power (relaxation) and modulation of pain-related potentials (P2, N2) after acupuncture — effects consistent with central analgesia.
fNIRS (near-infrared spectroscopy)
Measures superficial cortical oxygenation. Cheaper and more portable than fMRI. Recent studies confirm prefrontal activation consistent with effects on affective regulation and attention.
Limitations of the Research
Despite the volume of studies, the literature has acknowledged limitations:
- Most studies are in healthy volunteers, not in patients with the condition that acupuncture treats. The correlation between neuroimaging findings and clinical outcome is not always direct.
- Small sample sizes (mean of 15-30 subjects) and multiple statistical tests increase the risk of false-positive findings.
- Methodological heterogeneity: selected points, depth, manipulation, duration, and type of sham vary — making direct meta-analysis difficult.
- Imperfect blinding in studies with sham that involve real needling, even if superficial.
- The movement artifact introduced by needle manipulation is a technical acquisition challenge.
Clinical Implications
What does this literature mean for clinical practice?
- Acupuncture has documented central mechanisms that justify its use in conditions with a strong central component — chronic pain, fibromyalgia, migraine, anxiety, pain-associated depression.
- Deqi is not folklore: it correlates with the magnitude of the central effects. Performing the technique correctly matters.
- The clinical effect is biopsychosocial: it combines specific components (central modulation) and non-specific components (rapport, expectation, parasympathetic). Recognizing this improves communication with the patient, without reducing the technique to placebo.
- In peripheral conditions (localized myofascial pain, trigger point), the local segmental component dominates — and the fMRI literature is less relevant to clinical use.
Evidence behind this recommendation.
Selected studies from our library that inform the recommendations on this page. Evidence grade shown when available.
Modulatory effects of acupuncture on raphe nucleus-related brain circuits in patients with chronic neck pain
“A randomized fMRI study showed that verum acupuncture outperforms sham in chronic neck pain. Neurofunctional analysis revealed modulation of the raphe nucleus's serotonergic circuits — supporting a specific neurobiological basis for the observed clinical effect.”
Phantom Acupuncture Induces Placebo Credibility and Vicarious Sensations: A Parallel fMRI Study of Low Back Pain Patients
“Patients with low back pain showed measurable benefits under "phantom" (elaborate placebo) acupuncture. The authors do not invalidate real acupuncture — they separate contextual components (expectation, credibility) from specific components (neural stimulus from needling) in the clinical response.”
Myths and Facts
Myth vs. Fact
Acupuncture is just placebo — fMRI shows real and sham are the same.
Careful studies show consistent neural differences between verum and sham acupuncture — particularly in the PAG, hypothalamus, and posterior insula. Sham is not a neutral placebo: it is a mild form of real sensory stimulation.
The fMRI effects are too small to have clinical relevance.
Absolute BOLD effects are modest but reproducible and biologically coherent — they activate precisely the networks that process pain and emotion. The clinical effect's magnitude does not depend solely on the BOLD signal's magnitude.
Since it works at any point, classical channel theory has no modern anatomical support.
The literature is more nuanced. Somatotopic specificity is documented (different points activate different patterns), but partial overlap exists too. The traditional channel concept remains a useful clinical cartography, even if its anatomical interpretation is still debated.
Frequently Asked Questions
fMRI proves acupuncture has objective, reproducible neural effects — distinct from zero. Clinical efficacy in specific conditions is established by randomized trials, not directly by neuroimaging. The two evidence types complement each other.
To identify central mechanisms, refine point selection, predict clinical response, and separate specific from non-specific effect components. It also helps when communicating with patients and skeptical physicians, showing a measurable neural basis.
Task BOLD measures activation during the stimulus (while acupuncture is being applied). Resting-state measures sustained connectivity changes between brain regions before and after the stimulus, capturing longer-lasting effects — particularly useful for understanding why clinical effects persist between sessions.
No. It works best — and the neural evidence is most solid — in conditions with a strong central component (chronic pain, fibromyalgia, migraine, anxiety). In purely local conditions without chronic pain, central effects are less pronounced and the segmental/local component dominates.
Related Reading
Deepen your knowledge with related articles
Mechanisms of Acupuncture
An integrated view of local, segmental, and supraspinal mechanisms.
How Acupuncture Blocks Pain
The gate control theory and descending modulation.
Central Sensitization
Why chronic pain is neural reorganization — and how acupuncture acts on it.
Acupuncture, Sham, and Clinical Studies
How to interpret clinical evidence given sham-control limitations.