If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
Division of Neurosurgery, Toronto Western Hospital, University of Toronto, CanadaJoint Department of Medical Imaging, University of Toronto, Toronto, Canada
Division of Neurosurgery, Toronto Western Hospital, University of Toronto, CanadaDepartment of Neurosurgery, Tartu University Hospital, University of Tartu, Tartu, Estonia
Department of Neurosurgery, Henan University School of Medicine, Zhengzhou, ChinaDepartment of Neurosurgery, University of Louisville, Louisville, KY, United States
Edmond J. Safra Program in Parkinson's Disease and Morton and Gloria Shulman Movement Disorders Clinic, Toronto Western Hospital and Division of Neurology, UHN, Division of Neurology, University of Toronto, Toronto, Ontario, CanadaCenter for Advancing Neurotechnological Innovation to Application (CRANIA), Toronto, Ontario, Canada
fMRI is a powerful technique for observing the network-wide effects of DBS in vivo.
•
fMRI studies reveal that DBS modulates large-scale brain networks.
•
fMRI responses are related to stimulation site, stimulation parameters, patient characteristics, and clinical outcomes.
•
Recent studies have proposed fMRI-based biomarkers for confirming circuit engagement and determining treatment efficacy.
•
These findings highlight ways in which the technique could be used to refine DBS therapy.
Abstract
Background
Deep brain stimulation (DBS) is an established treatment for certain movement disorders and has additionally shown promise for various psychiatric, cognitive, and seizure disorders. However, the mechanisms through which stimulation exerts therapeutic effects are incompletely understood. A technique that may help to address this knowledge gap is functional magnetic resonance imaging (fMRI). This is a non-invasive imaging tool which permits the observation of DBS effects in vivo.
Objective
The objective of this review was to provide a comprehensive overview of studies in which fMRI during active DBS was performed, including studied disorders, stimulated brain regions, experimental designs, and the insights gleaned from stimulation-evoked fMRI responses.
Methods
We conducted a systematic review of published human studies in which fMRI was performed during active stimulation in DBS patients. The search was conducted using PubMED and MEDLINE.
Results
The rate of fMRI DBS studies is increasing over time, with 37 studies identified overall. The median number of DBS patients per study was 10 (range = 1–67, interquartile range = 11). Studies examined fMRI responses in various disease cohorts, including Parkinson's disease (24 studies), essential tremor (3 studies), epilepsy (3 studies), obsessive-compulsive disorder (2 studies), pain (2 studies), Tourette syndrome (1 study), major depressive disorder, anorexia, and bipolar disorder (1 study), and dementia with Lewy bodies (1 study). The most commonly stimulated brain region was the subthalamic nucleus (24 studies). Studies showed that DBS modulates large-scale brain networks, and that stimulation-evoked fMRI responses are related to the site of stimulation, stimulation parameters, patient characteristics, and therapeutic outcomes. Finally, a number of studies proposed fMRI-based biomarkers for DBS treatment, highlighting ways in which fMRI could be used to confirm circuit engagement and refine DBS therapy.
Conclusion
A review of the literature reflects an exciting and expanding field, showing that the combination of DBS and fMRI represents a uniquely powerful tool for simultaneously manipulating and observing neural circuitry. Future work should focus on relatively understudied disease cohorts and stimulated regions, while focusing on the prospective validation of putative fMRI-based biomarkers.
Deep brain stimulation (DBS) is a minimally invasive therapy in which surgically implanted electrodes deliver constant stimulation to precisely defined brain regions, thereby modulating aberrant neural circuitry and providing therapeutic benefit for various disorders [
]. DBS is a well-established treatment for movement disorders, including Parkinson's disease (PD), essential tremor (ET), and dystonia, and has shown promise for a growing list of psychiatric, cognitive, and seizure disorders [
To achieve therapeutic effects, DBS electrodes are precisely targeted to stimulate specific neural substrates implicated in the pathophysiology of various diseases. For example, in movement disorders, these targets include the subthalamic nucleus (STN), globus pallidus internus (GPi), and ventral intermediate nucleus of the thalamus (VIM), which are known to be critical hubs within the cortico-basal ganglia-thalamo-cortical circuit. Increasingly, however, there has been a shift in focus from what is being stimulated at the local level, toward visualising what is being engaged at the brain-wide, or network level [
], which have examined the structural and functional connectivity of stimulated brain regions and how these relate to clinical outcomes. However, because functional imaging is seldom acquired in DBS patients themselves, these studies have often leveraged ‘normative connectomes’. Normative connectomes serve as atlases of average brain connectivity calculated from large cohorts of subjects. Using normative connectomes, investigators can evaluate the connectivity of a given region in an average – or ‘normal – brain [
]. This is an important limitation, as a patient's brain connectivity may diverge from normal brain connectivity, particularly in the context of disease (e.g. Parkinson's disease) [
]. Furthermore, normative connectomes cannot show DBS-evoked brain responses; they can only say how a stimulated region is connected with the rest of the brain, not how stimulation actually affects neural activity [
In contrast to normative connectomic analyses, various techniques permit the patient-specific observation of DBS effects in vivo, including positron emission tomography (PET), single-photon emission computed tomography (SPECT), functional magnetic resonance imaging (fMRI), electroencephalography (EEG), and magnetoencephalography (MEG) [
Modulation of CNS functions by deep brain stimulation: insights provided by molecular imaging.
in: Dierckx R.A.J.O. Otte A. de Vries E.F.J. van Waarde A. Leenders K.L. PET and SPECT in neurology, cham: springer international publishing. 2021: 1177-1244
]. Of these modalities, fMRI – which evaluates brain activity through fluctuations in blood-oxygen-level-dependent (BOLD) signal – represents somewhat of a compromise, with better temporal resolution than PET and SPECT, and superior spatial resolution compared to EEG and MEG. Moreover, BOLD is an endogenous contrast that does not require the ingestion or injection of foreign material. These factors mean that fMRI can be used for various experimental paradigms, including the assessment of neural activity at rest, or brain responses to a given task or stimulus [
]. The former paradigm in particular allows for the study of functional connectivity between distant brain regions, as measured by BOLD signal coherence [
Deep brain stimulation initiative: toward innovative technology, new disease indications, and approaches to current and future clinical challenges in neuromodulation therapy.
Deep brain stimulation initiative: toward innovative technology, new disease indications, and approaches to current and future clinical challenges in neuromodulation therapy.
]. Findings from fMRI DBS studies are an invaluable addition to the connectomic cannon and demonstrate ways in which the technique could be leveraged to establish imaging-based biomarkers, confirm network engagement, and identify new targets for DBS. Accordingly, in this review, our objectives were to i) identify studies from the existing fMRI DBS literature, ii) describe pertinent study and subject characteristics, iii) summarize key findings, iv) outline common technical challenges, v) uncover gaps in the literature, and vi) highlight emerging trends. Herein, we present an exciting and expanding field, showing that the combination of DBS and fMRI represents a uniquely powerful tool for simultaneously manipulating and observing neural circuitry.
2. Methods
We conducted a systematic review of all published, peer-reviewed original research pertaining to DBS and fMRI studies in humans. This review was conducted on 2nd February 2022 by searching the MEDLINE and NCBI PubMED databases using the exploded search terms “DBS” AND “fMRI”. The search generated a combined total of 4866 results. All relevant articles were selected for full-text review and had to meet the following inclusion and exclusion criteria: i) inclusion (articles reporting fMRI responses to active stimulation in humans), and ii) exclusion (animal studies; reviews; conference abstracts; studies lacking neuroimaging findings; studies that did not scan patients with active stimulation). After removing duplicate entries (n = 1696), the remaining studies were screened according to title and abstract by two independent reviewers (C.T.C., D.G.) through Covidence (https://www.covidence.org/). A third rater (A.L.) resolved any disagreements. After removing studies based on title and abstract screening (n = 2894), 46 studies were left for full text screening. At this stage, reviews (n = 2), animal studies (n = 3), normative neuroimaging studies, conference abstracts (n = 1), safety studies lacking neuroimaging findings, studies that did not use fMRI (n = 3), and fMRI studies that did not scan patients with DBS turned on (n = 2) were excluded. Two additional studies were identified through citation searching. As described in PRISMA-S, citation searching is a method of identifying eligible studies by examining cited and citing references [
]. Of the remaining 37 studies, information regarding sample size, clinical indication and demographics, stimulation target, stimulation parameters, and functional neuroimaging changes were extracted. This process is summarized in Fig. 1.
Fig. 1PRISMA flow diagram of the screening process. This figure presents the process of identifying eligible articles to include in the systematic review, including the initial search (‘Identification’) and the subsequent vetting process (‘Screening’). The number of studies excluded at each step and reasons for exclusion are specified. Note: This PRISMA figure was adapted from Page et al. [
We identified 37 studies that satisfied the inclusion criteria (Fig. 1). These studies are summarized in Table 1 and Table 2. Highlighting a growing interest within the field, the rate at which fMRI DBS studies are being published is increasing over time (Fig. 2), with studies encompassing a wide range of disorders (i.e. DBS indications), experimental designs, stimulated brain regions, MRI hardware, and cohort size (Fig. 3). Findings from each DBS indication and a summary of experimental designs used are expanded below. On average, studies included 13 (±14) (median = 10, range: 1–67, interquartile range [IQR] = 11.5) DBS participants, with 5 studies additionally scanning non-DBS control subjects. While an increasing number of these studies are being performed at 3-T (T) field strength, the majority have been performed at 1.5T (Fig. 2).
Table 1A tabular summary of all included fMRI DBS studies. Pertinent study characteristics are shown, including patient disease, number of subjects, DBS target, MRI hardware, and fMRI paradigm. Finally, the key findings from each study are briefly described.
Thalamic stimulation and functional magnetic resonance imaging: localization of cortical and subcortical activation with implanted electrodes. Technical note.
Resting-state functional magnetic resonance imaging of the subthalamic microlesion and stimulation effects in Parkinson's disease: indications of a principal role of the brainstem.
Penetration of electrodes was associated with increased EC of functional connectivity in the brainstem and changes in connectivity were quantitatively related to motor improvement.
Motor and nonmotor circuitry activation induced by subthalamic nucleus deep brain stimulation in patients with Parkinson disease: intraoperative functional magnetic resonance imaging for deep brain stimulation.
Functional connectivity between the somatosensory-motor cortices and thalamus, and between the somatosensory-motor cortices and cerebellum decreased with STN- and GPi-DBS
DBS ON/OFF x food conditions (sweet vs savory vs neutral foods)
IN, SMG, Th, Am, PF
STN-DBS was associated with increased functional connectivity within the salience network and decreased activity within the reward-related network in the context of sweet food images.
Deep brain stimulation for obsessive-compulsive disorder: using functional magnetic resonance imaging and electrophysiological techniques: technical case report.
Intraoperative functional magnetic resonance imaging for monitoring the effect of deep brain stimulation in patients with obsessive-compulsive disorder.
Thalamic stimulation and functional magnetic resonance imaging: localization of cortical and subcortical activation with implanted electrodes. Technical note.
Acute brain activation patterns of high- versus low-frequency stimulation of the anterior nucleus of the thalamus during deep brain stimulation for epilepsy.
Motor and nonmotor circuitry activation induced by subthalamic nucleus deep brain stimulation in patients with Parkinson disease: intraoperative functional magnetic resonance imaging for deep brain stimulation.
Resting-state functional magnetic resonance imaging of the subthalamic microlesion and stimulation effects in Parkinson's disease: indications of a principal role of the brainstem.
Thalamic stimulation and functional magnetic resonance imaging: localization of cortical and subcortical activation with implanted electrodes. Technical note.
Deep brain stimulation for obsessive-compulsive disorder: using functional magnetic resonance imaging and electrophysiological techniques: technical case report.
Acute brain activation patterns of high- versus low-frequency stimulation of the anterior nucleus of the thalamus during deep brain stimulation for epilepsy.
Studies are organised by DBS target. Common findings and notable differences were only identified if there were at least two studies of the same target and experimental design. Comparisons were not made between experimental designs. Abbreviations: ALIC = anterior limb of the internal capsule; ANT = anterior nucleus of the thalamus; BOLD = blood-oxygen-level-dependent signal; CM-pf = centromedian parafascicular complex; CZI = caudal zona incerta; DBS = deep brain stimulation; DLB = dementia with Lewy bodies; GPi = globus pallidus internus; M1 = primary motor cortex; PVG = periventricular grey; SCC = subcallosal cingulate cortex; SMA = supplementary motor area; STN = subthalamic nucleus; VIM = ventral intermediate nucleus of the thalamus; Vc = nucleus ventrocaudalis; VC/VS = ventral capsule/ventral striatum.
Fig. 2Cumulative number of fMRI DBS studies over time. The overall rate of DBS fMRI studies is increasing over time. The rate of all DBS fMRI studies is shown in blue, while studies performed with 1.5T MRI are shown in red and studies performed with 3T MRI are shown in yellow. Abbreviations: DBS = deep brain stimulation; fMRI = functional magnetic resonance imaging; No. = number; T = tesla.
Fig. 3Study cohorts, experimental paradigms, and stimulated brain regions. A breakdown of fMRI DBS studies by disease cohort (A), experimental paradigms (B), and stimulated brain regions (C). For experimental paradigms, cs-fMRI denotes studies in which in which the patient is at rest and DBS is periodically turned ON and OFF (i.e. stimulation itself is the task), tb-fMRI denotes studies in which the participant periodically performs a given task (e.g. finger tapping) while the DBS condition remains constant, and rs-fMRI denotes studies in which the patient is at rest and the DBS condition remains constant. Abbreviations: ALIC = anterior limb of the internal capsule; ANT = anterior nucleus of the thalamus; cs-fMRI = cycling-stimulation fMRI; DBS = deep brain stimulation; ET = essential tremor; fMRI = functional magnetic resonance imaging; GPi = internal globus pallidus; OCD = obsessive compulsive disorder; PD = Parkinson's disease; rs-fMRI = resting state fMRI; STN = subthalamic nucleus; tb-fMRI = task-based fMRI; VIM = ventral intermediate nucleus of the thalamus; VC/VS = ventral capsule/ventral striatum.
Broadly speaking, fMRI DBS acquisitions could be categorized according to three different experimental designs: i) resting-state fMRI (rs-fMRI), in which the patient is at rest and the DBS condition remains constant (e.g. DBS turned ON or OFF), ii) cycling-stimulation fMRI (cs-fMRI), in which the patient is at rest and DBS is periodically turned ON and OFF, or iii) task-based fMRI (tb-fMRI), in which the participant periodically performs a given task (e.g. finger tapping) while the DBS condition remains constant. These designs were leveraged to explore the impact of DBS on brain networks at rest (i and ii), or how DBS modulates brain activity during simple tasks or external stimuli (iii). These different acquisitions also facilitated different types of analyses, such that rs-fMRI acquisitions were most commonly used to study functional connectivity, while task/simulation-based studies showed areas of relatively increased or decreased BOLD activation associated with that task. We identified 10 studies employing rs-fMRI (i), 21 studies using cs-fMRI (ii), and 11 studies using tb-fMRI (iii).
3.2 DBS indications
Included studies examined the effects of DBS in various patient populations, including movement, psychiatric, cognitive, and seizure disorders (Fig. 3A).
3.2.1 Parkinson's disease
Commensurate with the fact that PD is the most common indication for DBS worldwide, two thirds (n = 24) of all included studies examined PD patients. The majority of patients received DBS targeting the subthalamic nucleus (STN), with only a handful of studies reporting the effects of globus pallidus internus (GPi) stimulation (n = 3). These studies used all three of the aforementioned experimental designs to glean mechanistic and translational insights into the effects of DBS on neural circuitry in PD patients.
Most commonly, PD studies employed a cs-fMRI design (n = 10). By cycling stimulation ON and OFF with the patient at rest, these studies clearly and intuitively delineated brain regions where stimulation alone was associated with increased or decreased BOLD activation [
Motor and nonmotor circuitry activation induced by subthalamic nucleus deep brain stimulation in patients with Parkinson disease: intraoperative functional magnetic resonance imaging for deep brain stimulation.
]. All groups examined the effects of subthalamic stimulation, with Boutet et al. and Dimarzio et al. additionally investigating a smaller cohort of patients with pallidal stimulation. Across these studies, subthalamic stimulation was consistently associated with significant BOLD changes within the cortico-basal ganglia-thalamo-cortical loop and cerebellum. While these brain regions were reliably engaged, there were differences in whether stimulation was associated with increased or decreased BOLD activation. Indeed, subthalamic stimulation was associated with increased BOLD activation in the primary motor cortex in n = 5 studies, while two of the largest such studies demonstrated significantly decreased activation in this region [
]. The effects of subthalamic stimulation ON > OFF across all 10 cs-fMRI studies are summarized in Fig. 4.
Fig. 4BOLD responses associated with DBS ON > OFF in PD patients. Common areas of significant BOLD activation associated with subthalamic DBS ON > OFF across all 10 cs-fMRI PD studies are shown. Areas of significantly decreased BOLD activity are shown in cool colours, while areas of significantly increased BOLD activity are shown in warm colours. Studies commonly implicated cerebellum, basal ganglia, and sensorimotor cortex. The methods for creating this figure are described in the supplementary materials. Abbreviations: BOLD = blood-oxygen-level-dependent signal; cs-fMRI = cycling-stimulation functional magnetic resonance imaging; DBS = deep brain stimulation; PD = Parkinson's disease.
Complementing findings from cs-fMRI acquisitions, several studies (n = 8) employed rs-fMRI to examine the effects of DBS on functional connectivity in PD patients. These studies consistently demonstrated that DBS is associated with an increase in thalamo-cortical connectivity [
], while Holiga and colleagues showed that STN-DBS in 13 patients was associated with a greater quantity and strength of connections of the premotor cortex [
Resting-state functional magnetic resonance imaging of the subthalamic microlesion and stimulation effects in Parkinson's disease: indications of a principal role of the brainstem.
]. Conversely, Kahan and colleagues showed that stimulation was associated with decreased STN connectivity, including striatal-STN, STN-thalamic, and cortico-STN (hyperdirect) connectivity [
]. Of note, two studies using healthy controls demonstrated that STN-DBS normalizes functional connectivity in PD patients towards healthy brain states [
Studies have also investigated how DBS modulates neural activity associated with motor tasks. An early study by Hesselman and colleagues showed, during a left-sided finger opposition task, a decrease in BOLD signal activation when DBS was ON compared to OFF in the contralateral sensorimotor cortex and ipsilateral cerebellum [
]. Kahan and colleagues additionally showed stimulation evoked modulations in cortico-cerebellar connectivity during voluntary movement, which were not present with STN-DBS at rest [
], suggesting that DBS has both behavior-dependent and behavior-independent effects on motor circuitry.
Many of these studies have also explored the relationship between stimulation-evoked fMRI responses, patient characteristics, stimulation parameters, clinical outcomes, and DBS target.
Examining the influence of PD subtypes in a cs-fMRI study, Dimarzio and colleagues showed variable degrees of response in primary and secondary motor areas depending on whether patients had tremor dominant, akinesia-rigidity dominant, or postural instability gait disorder subtypes [
]. In a recent longitudinal study with a cs-fMRI design, Shen and colleagues revealed time-dependent effects of subthalamic stimulation, showing decreased primary motor cortex (M1) BOLD activation over time [
]. The same study additionally revealed that a distinct pallido-thalamo-cerebellar circuit was preferentially modulated by high frequency, rather than low frequency stimulation [
]. Furthermore, in another cs-fMRI study, Hancu and colleagues demonstrated the differential effects of stimulation configuration on BOLD response (i.e. monopolar versus bipolar stimulation), showing that clinically optimal monopolar stimulation was associated with greater BOLD activation in the thalamus and pallidum than the equivalent bipolar configuration [
]. Reinforcing these findings, in the largest fMRI DBS study to date (n = 67), Boutet and colleagues used cs-fMRI to demonstrate that clinically optimal stimulation parameters - irrespective of active electrode contact, voltage, and frequency - were associated with a signature of BOLD response involving deactivation of M1 and cerebellum, and activation of the thalamus. This signature was predictive of patients’ clinically optimal stimulation parameters in unseen datasets, highlighting the potential of fMRI to be used in the selection of patient-specific DBS settings [
]. Additionally, studies have directly examined the relationship between BOLD responses and therapeutic or adverse clinical outcomes from DBS. Using intraoperative cs-fMRI, Gibson and colleagues showed that improvement in contralateral tremor was correlated with activation of thalamic, brainstem, and cerebellar regions, while improvements in bradykinesia and rigidity were most strongly correlated with activation in the primary motor cortex. Conversely, the presence of unwanted muscle contractions was correlated with striatal activation, while nausea and paresthesias were related to sensorimotor activation [
]. Finally, studies have compared the differing effects of GPi- and STN-DBS. In a cs-fMRI study, Dimarzio et al. found that GPi-DBS (n = 3 subjects) was associated with significantly more primary motor cortex activation than STN-DBS (n = 4 subjects) [
]. Further, a recent rs-fMRI study found that while STN-DBS and GPi-DBS were both associated with significant changes in thalamo-cortical and cortico-cerebellar functional connectivity, STN-DBS was associated with more widespread functional connectivity changes than GPi-DBS, including between somato-motor and visual regions [
fMRI has additionally been used to explore the effects of DBS in other movement disorders, including ET and tic disorders.
We identified two fMRI studies investigating ET patients receiving DBS of the ventral intermediate (VIM) nucleus of the thalamus. Gibson et al. and Rezai et al. employed a cs-fMRI design, demonstrating that, similar to STN-DBS, VIM-DBS was associated with ipsilateral thalamic and sensorimotor cortex activation [
Thalamic stimulation and functional magnetic resonance imaging: localization of cortical and subcortical activation with implanted electrodes. Technical note.
]. Gibson et al. additionally demonstrated significant BOLD activation in the contralateral cerebellar cortex and deep cerebellar nuclei. Stimulation-evoked activation in all of these regions was significantly correlated with the therapeutic effectiveness of DBS, with cerebellar activation exhibiting the strongest correlations [
DBS of the caudal zona incerta (cZI) - a slip of grey matter between the thalamic and lenticular fasiculi - has also gained traction as a target for the treatment of tremor. In a recent study, Awad and colleagues employed two experimental designs, incorporating tremor-inducing tasks (tb-fMRI), as well as rs-fMRI in 16 cZI-DBS subjects with ET. They showed that cZI-DBS was associated with task-dependent and task-independent modulation of sensorimotor cerebello-cerebral regions, highlighting the complexity of stimulation mechanisms [
]. Task-independent effects (i.e. main effect of DBS) were increased premotor activation during motor tasks and rest, whereas, task-dependent effects were decreased activation of sensorimotor cortex and cerebellar lobule VIII during the holding task, with increased activation of SMA and cerebellar lobule V during rest.
DBS has also demonstrated effectiveness in tic disorders, characterized by sudden and repetitive movements or sounds. In an cs-fMRI study of five patients with Tourette syndrome, stimulation of the centromedian parafascicular complex (CM-pf) of the thalamus was associated with significant BOLD changes in prefrontal cortex, parietal cortex, sensorimotor cortex, cerebellum, thalamus, basal ganglia, and limbic areas [
]. Notably, suppression of motor and insula networks was correlated with motor tic reduction, while suppression of frontal and parietal networks correlated with vocal tic reduction.
3.2.3 Psychiatric and cognitive disorders
DBS has increasingly been trialed for psychiatric and cognitive indications with varying degrees of success. A greater understanding of how DBS engages psychiatric and cognitive networks is therefore necessary to optimize therapy and refine stimulation targets. DBS for the treatment of medically refractory obsessive-compulsive disorder (OCD) has been FDA approved since 2009, with groups targeting various substrates, including the ventral capsule/ventral striatum (VC/VS), nucleus accumbens (NAcc), STN, inferior thalamic peduncle, and bed nucleus of the stria terminalis [
]. To date, two fMRI studies have explored the effects of VC/VS-DBS, with one study investigating NAcc-DBS. These studies leveraged cs-fMRI paradigms, with all three demonstrating ipsilateral striatal activation [
Intraoperative functional magnetic resonance imaging for monitoring the effect of deep brain stimulation in patients with obsessive-compulsive disorder.
Deep brain stimulation for obsessive-compulsive disorder: using functional magnetic resonance imaging and electrophysiological techniques: technical case report.
Deep brain stimulation for obsessive-compulsive disorder: using functional magnetic resonance imaging and electrophysiological techniques: technical case report.
Intraoperative functional magnetic resonance imaging for monitoring the effect of deep brain stimulation in patients with obsessive-compulsive disorder.
]. Gibson and colleagues additionally showed that stimulation-evoked smiling and laughter - a well-documented side effect of VC/VS-DBS - was associated with decreased activation of the anterior cingulate cortex. Studies have yet to show an association between stimulation induced fMRI responses and symptom improvement in OCD DBS patients.
DBS of the subcallosal cingulate cortex (SCC) is a well-tolerated investigatory treatment for major depressive disorder (MDD), bipolar disorder, and anorexia nervosa [
]. In a recent rs-fMRI study, Elias et al. explored fMRI responses to SCC-DBS in 16 subjects with MDD (n = 6), BD (n = 1), or AN (n = 9). Stimulation was associated with increased activity in dorsal anterior cingulate cortex (dACC), posterior cingulate cortex (PCC), precuneus, and left inferior parietal lobule. Further, stimulation-induced dACC signal reduction correlated with immediate within-session mood fluctuations and was greater at a priori clinically determined optimal versus suboptimal settings. Moreover, immediate changes in dACC, PCC, and precuneus activity were predictive of individual long-term antidepressant improvement [
DBS has also been evaluated for cognitive disorders. In an rs-fMRI study of six patients with dementia with Lewy bodies (DLB), stimulation of the nucleus basalis of Meynert (NBM) was associated with functional connectivity changes in the default mode network and frontoparietal network [
]. Specifically, DBS was associated with decreased connectivity between the posterior cingulate cortex and the right inferior parietal lobule, with increased connectivity between the left intraparietal sulcus and the left inferior frontal gyrus, and the left superior parietal lobule (precuneus), and the right paracingulate gyrus. However, there was no reported association between these network changes and improvement in cognitive function.
3.2.4 Epilepsy
There is a long tradition investigating the use of brain stimulation in the treatment of seizures, and the efficacy of DBS for refractory epilepsy has been evaluated with multiple randomized trials [
Seizure outcome after hippocampal deep brain stimulation in patients with refractory temporal lobe epilepsy: a prospective, controlled, randomized, double-blind study.
]. Based on the findings of the SANTE trial, stimulation of anterior thalamic nucleus (ANT) of the thalamus received US Food and Drug Administration (FDA) approval in the United States in 2018 [
Acute brain activation patterns of high- versus low-frequency stimulation of the anterior nucleus of the thalamus during deep brain stimulation for epilepsy.
]. ANT stimulation produced activation within several areas of the brain corresponding to the default mode and limbic networks, including the thalamus, bilateral anterior cingulate and posterior cingulate cortex, precuneus, medial prefrontal cortex, amygdala, ventral tegmental area, hippocampus, striatum, and right angular gyrus. Furthermore, Sarica et al. and Middlebrooks et al. showed that these areas were not activated when the locus of stimulation was moved away from the ANT and when low-frequency stimulation (30 Hz) was used, further highlighting the potential for fMRI in selecting optimal stimulation parameters [
Acute brain activation patterns of high- versus low-frequency stimulation of the anterior nucleus of the thalamus during deep brain stimulation for epilepsy.
]. In one of the first fMRI DBS studies, Rezai and colleagues used a cs-fMRI design to investigate a patient with leads targeting the ventrocaudalis nucleus (Vc) of the thalamus and periventricular grey area (PVG) [
Thalamic stimulation and functional magnetic resonance imaging: localization of cortical and subcortical activation with implanted electrodes. Technical note.
]. Vc-DBS intensities provoking paresthesias resulted in activation of the primary somatosensory cortex (SI), thalamus, secondary somatosensory cortex (SII), and insula. In contrast, PVG stimulation did not activate the SI, but resulted in medial thalamic and cingulate cortex activation, potentially illustrating different network effects between the two targets.
In a more recent tb-fMRI study, Jones and colleagues examined how DBS of the anterior limb of the internal capsule (ALIC) modulated pain in five patients with central post-stroke pain [
]. In response to noxious stimuli, patients in the DBS OFF state showed significant activation in the thalamus, insula, operculum, and orbitofrontal regions. ALIC-DBS significantly reduced the activation of these regions. Noxious stimuli during ALIC-DBS were additionally associated with decreased activation in the hippocampi.
4. Discussion
A review of the fMRI DBS literature reveals a small, yet rapidly growing field. We identified 37 studies to date, over half of which have been published in the last 5 years. The majority of studies have investigated STN-DBS in PD. There was a notable paucity of GPi-DBS studies, particularly when considering the established clinical use of this target in PD and dystonia. The remainder of fMRI DBS studies evaluated the effects of stimulation in emerging and experimental indications, including psychiatric, cognitive, seizure, and pain disorders. Studies showed that DBS modulates large-scale brain networks, and that stimulation-evoked fMRI responses are related to the site of stimulation, stimulation parameters, patient characteristics, and therapeutic outcomes. Recent studies have also proposed fMRI-based biomarkers for DBS treatment, highlighting ways in which fMRI could be used to confirm circuit engagement and refine DBS therapy.
4.1 Limitations
We identified several limitations across the fMRI DBS literature that should be taken into account when considering findings and when planning future studies. Firstly, the median DBS participant cohort size was relatively small (10), with 3 studies (PD: n = 2; OCD: n = 1) including a single patient. Indeed, when excluding non-PD indications, the average sample size was 6.1 ± 4.9 DBS patients. This is comparable to the sample sizes of the 1000 most highly-cited fMRI studies between 1990 and 2012 (n = 12), but less than the average sample size of fMRI studies published in high impact neuroimaging journals (Nature Neuroscience, The Journal of Neuroscience, NeuroImage and Cerebral Cortex) in 2017/18 (average = 24) [
Sample size evolution in neuroimaging research: an evaluation of highly-cited studies (1990-2012) and of latest practices (2017-2018) in high-impact journals.
]. This reflects the logistical difficulties of scanning DBS patients, including a relatively small patient pool (approximately 200,000 people have undergone DBS worldwide) and MRI safety considerations. Despite these challenges, fMRI DBS studies have been growing larger, with a recent study including over 60 DBS participants with PD [
Secondly, while the array of experimental designs and analyses available to fMRI investigators is an undoubted advantage of the technique, the comparability of findings across studies is challenging. This is particularly the case when comparing rs-fMRI studies (examining changes in functional connectivity) and task-based studies (examining increases/decreases in BOLD response). Future studies could utilize multiple fMRI metrics in the same participants to better understand the relationship between stimulation-evoked functional connectivity changes and BOLD response.
Even when studies were directly comparable, there were inconsistencies in findings. This can be seen in Fig. 4, in which 5 studies (comprising 47 subjects) showed increased sensorimotor cortex BOLD response with STN-DBS, while 2 large studies (comprising 89 subjects) showed decreased response in the same area. These seemingly contradictory results may be a function of noise, particularly when considering differences in MRI hardware, stimulation parameters, subject characteristics, and image preprocessing between studies. Inconsistencies may also be related to a more general limitation of fMRI, in that it is a proxy - rather than a direct - measure of neuronal activity that reflects a complex interplay between cerebral blood flow, cerebral blood volume, and the cerebral metabolic rate of oxygen [
Intraoperative functional magnetic resonance imaging for monitoring the effect of deep brain stimulation in patients with obsessive-compulsive disorder.
]. This is particularly worth noting in pathological or aged brain states, in which altered neurovascular coupling may complicate interpretation of BOLD response (e.g. decreased BOLD signal may be associated with increased neuronal activity [
Deep brain stimulation for obsessive-compulsive disorder: using functional magnetic resonance imaging and electrophysiological techniques: technical case report.
]). Finally, discordant findings may also reflect a stimulation-clinical response confound, which is an inherent and infrequently addressed feature of fMRI DBS studies (this confound is discussed further below).
Owing to these limitations, fMRI alone – at least with conventional analyses and MRI hardware – may be insufficient in developing robust physiological models of stimulation effects. To this end, studies could instead integrate fMRI with modalities possessing superior temporal resolution, such as EEG and MEG, as well as tools that permit more granular inspection of neuronal activity, such as microelectrode recordings (MER). For example, fMRI could be used for broadly identifying network hubs associated with stimulation that could then be interrogated further with EEG, MEG, or MER.
These limitations are less relevant when the primary goal is to isolate a clinically useful biomarker of stimulation. In this sense, the most important feature is that the fMRI biomarker – whether it be a direct or indirect measure of stimulation – is tightly linked to the clinical outcome of interest. Indeed, an unsupervised machine learning algorithm may be trained to predict DBS clinical outcomes based on BOLD responses with 100% accuracy, while not necessarily offering pure neuroscientific or physiological insights.
4.2 The stimulation-clinical response confound
To gain a clearer understanding of DBS effects, studies should give consideration to whether BOLD responses are: i) due to the direct effects of stimulation on the brain, ii) a reflection of the clinical effects of stimulation, whether they be physical (e.g. reduced tremor), cognitive, or affective, iii) or a combination of the two. There are various strategies that investigators could conceivably use to answer this question, though each have their own limitations.
Due to the latency of the haemodynamic BOLD response and the limited temporal resolution of fMRI (∼1–3 s), it is challenging to tease apart these effects based on chronology. An example of this is in PD, where the time between stimulation onset and symptom change can be almost immediate. It is also difficult to disentangle these effects when the clinical response to stimulation is not readily measurable. For example, in DBS for depression, it would be almost impossible to gauge exactly when a patient had a change in affect following the onset of stimulation.
For studies in which the goal is to isolate the direct effects of stimulation on BOLD response, a more practical alternative may be to try regress out, or completely nullify clinical responses to stimulation. This can be achieved when subjects are under general anesthesia, rendering intraoperative fMRI a valuable opportunity for answering mechanistic questions [
Motor and nonmotor circuitry activation induced by subthalamic nucleus deep brain stimulation in patients with Parkinson disease: intraoperative functional magnetic resonance imaging for deep brain stimulation.
]. However, it should also be noted that general anesthesia is a considerable confound in and of itself, with widespread effects on brain activity and BOLD signal [
]. For these reasons, comparisons between awake and anesthesia-state fMRI DBS studies should be made with caution.
4.3 Technical challenges
Many studies highlighted a number of the ongoing technical challenges associated with fMRI DBS, including MRI safety, hardware-related susceptibility artefact, and electromagnetic interference.
The prevailing challenge for fMRI DBS has been MRI safety, particularly concerns regarding radiofrequency induced heating around the DBS lead tips [
]. In order to stay within DBS-vendor guidelines, studies have typically been restricted to scanning at lower field strengths, which is reflected in the fact that 85% (n = 28) of included studies were performed at 1.5T. However, a growing understanding of MRI DBS safety has led to an increasing number of studies using 3T machines, allowing for improved signal-to-noise ratio (SNR) and optimized sequences for maximal temporal and spatial resolution (Fig. 2). These studies have gleaned a number of novel findings, such as the time and frequency dependent effects of STN-DBS, as well fMRI signatures that can be used to predict clinically optimal stimulation settings [
]. The boundaries of DBS fMRI studies could be pushed yet further, with a recent study showing preliminary evidence for the safety and improved imaging quality of 7T MRI in an ex vivo DBS model [
Another consideration for studies was the artefact from DBS leads (in the brain parenchyma) and extension wire (typically coiled under one side of the scalp). This artefact is created by the magnetic susceptibility of certain materials and causes signal dropout in the area around the hardware. Unfortunately, this means that signal loss can encompass relevant network hubs, including the region in the immediate vicinity of the target (e.g. STN), as well as cortical areas (e.g primary motor cortex). To account for these artefacts, studies have typically avoided analyzing implicated regions altogether by performing unilateral stimulation only (contralateral to the coil of excess of extension wire), or by segmenting and masking areas of signal loss. Notably, one study hinted at the potential utility of including regions of artefact in analyses, showing BOLD signal coherence between the areas immediately surrounding bilateral electrodes [
]. Future studies should investigate the extent to which these areas of artefact can be used in analyses, if at all.
Finally, studies have contended with concerns regarding electromagnetic interference (EMI) during fMRI acquisitions (i.e. the interaction of radiofrequency pulses with the electrical current from DBS). For this reason, a number of studies have elected to use bipolar, rather than monopolar stimulation (which is more clinically used), as bipolar stimulation is theoretically less susceptible to EMI. However, recent phantom studies have shown minimal EMI from monopolar stimulation [
]. These findings suggest that studies should scan with the most clinically-relevant configurations, including monopolar stimulation, when it is safe to do so.
4.4 Summary and future directions
Moving beyond a gross examination of brain responses to stimulation, studies are increasingly investigating the relationship between fMRI responses and clinical outcomes. While these studies have largely been restricted to PD patients, they hint at the vast translational potential of fMRI in DBS, including the identification of putative biomarkers for determining network engagement and optimal therapeutic response to stimulation. Such biomarkers could be used to guide the selection of stimulation parameters in DBS patients, which is currently a time-consuming process based on the subjective evaluation of a patient's clinical response to parameter changes [
]. Further, intraoperative fMRI responses to stimulation have been correlated with long-term improvement from DBS, suggesting that fMRI could be used to refine electrode placement or titrate stimulation parameters intraoperatively. Future fMRI DBS studies should focus on disorders such as dystonia, cognitive disorders, depression, and epilepsy, where clinical benefits from stimulation may take weeks or months to manifest, making immediate biomarkers of network engagement particularly valuable. Newer studies should also be able to take advantage of emerging technologies, such as DBS systems capable of local field potential (LFP) recordings, to develop multimodal biomarkers [
]. The combination of LFPs and fMRI in particular would provide rich and complementary data that could examine response to DBS with excellent temporal and spatial resolution. Finally, future studies should incorporate fMRI DBS into clinical trials to prospectively validate the technique as a tool for refining DBS therapy.
R.M. is an employee at General Electric. A.F. serves as a consultant for Medtronic, Abbott, Boston Scientific, Brainlab, Ceregate, and Medtronic, he received research grants, personal fees and non-financial support from Boston Scientific , Brainlab and Medtronic and personal fees from Abbott and Ceregate, all outside the submitted work. A.M.L. serves as a consultant for Medtronic, Abbott, Boston Scientific, and Functional Neuromodulation. The other authors report no conflict of interest concerning the materials or methods used in this study or the findings specified in this paper.
Appendix A. Supplementary data
The following are the supplementary data related to this article:
Modulation of CNS functions by deep brain stimulation: insights provided by molecular imaging.
in: Dierckx R.A.J.O. Otte A. de Vries E.F.J. van Waarde A. Leenders K.L. PET and SPECT in neurology, cham: springer international publishing. 2021: 1177-1244
Deep brain stimulation initiative: toward innovative technology, new disease indications, and approaches to current and future clinical challenges in neuromodulation therapy.
Motor and nonmotor circuitry activation induced by subthalamic nucleus deep brain stimulation in patients with Parkinson disease: intraoperative functional magnetic resonance imaging for deep brain stimulation.
Resting-state functional magnetic resonance imaging of the subthalamic microlesion and stimulation effects in Parkinson's disease: indications of a principal role of the brainstem.
Thalamic stimulation and functional magnetic resonance imaging: localization of cortical and subcortical activation with implanted electrodes. Technical note.
Intraoperative functional magnetic resonance imaging for monitoring the effect of deep brain stimulation in patients with obsessive-compulsive disorder.
Deep brain stimulation for obsessive-compulsive disorder: using functional magnetic resonance imaging and electrophysiological techniques: technical case report.
Seizure outcome after hippocampal deep brain stimulation in patients with refractory temporal lobe epilepsy: a prospective, controlled, randomized, double-blind study.
Acute brain activation patterns of high- versus low-frequency stimulation of the anterior nucleus of the thalamus during deep brain stimulation for epilepsy.
Sample size evolution in neuroimaging research: an evaluation of highly-cited studies (1990-2012) and of latest practices (2017-2018) in high-impact journals.
00060-2&id=gr1.jpg){kind=link}
00060-2&id=gr2.jpg){kind=link}
00060-2&id=gr3.jpg){kind=link}
00060-2&id=gr4.jpg){kind=link}