Transcranial Magnetic Stimulation during Gait: A Review of Methodological and Technological Challenges
Keywords: Gait, leg motor cortex, motor threshold, transcranial magnetic stimulation
Transcranial magnetic stimulation (TMS), also termed as a “perturbation” technique, is unique in establishing a causal link and functional connectivity between the brain areas and task performance. TMS is a noninvasive indirect application of an electrical current to temporarily disrupt local electric flow in a neuronal pool for a small duration of time, ~20 ms. The stimulation is delivered by passing 200 ms electrical current at 10,000 volts and 600 amps through a coil of wire called stimulator or coil., The current through the stimulator is termed as the primary current that induces a strong magnetic field (1-2 Tesla) perpendicular to the coil which in turn induces secondary electrical current in the brain tissue (eddy currents). The induced electric field is perpendicular to the magnetic field and parallel to the coil but in the opposite direction. When the induced current reaches a threshold intensity it results in synchronized depolarization of neurons within the focus of stimulation called neuronal “firing.” When stimulated over the primary motor cortex, the depolarization generates electrical potentials in the represented muscles and recorded as motor evoked potentials (MEP). The TMS has good temporal resolution in the order of 1 ms and spatial resolution of 1.5 cm with a standard figure-of-8 coil.,, TMS is widely used for therapeutic and research purposes and can be applied as a single, paired, or repetitive pulse.,,,
The objective of this narrative review is to present the sources of variations in research with TMS focusing specifically on the challenges involving TMS application to gait research. These challenges are listed as follows:
Placebo control: A placebo is defined as a substance or procedure that is without specific activity for the condition being evaluated. There are a few issues when it comes to setting-up an ideal placebo condition when using TMS. The application of TMS involves sensory perceptions that may interfere with task performance. These sensations are the clicking sound from the coil when stimulus is delivered, tactile sensation of coil over scalp, and visible muscle twitch when stimulating the motor cortex. A clicking sound is produced by the coil with each pulse delivery and proportionally increases with the stimulus frequency and intensity. A low-intensity sound may not draw the participant's attention and possibly less interfering with the task performance compared to a high-intensity sound. Additionally, sound by itself may modulate arousal and increase awareness of the coil's presence over the head. In case a placebo is part of a testing paradigm, the placebo set-up should provide an identical experience without any active stimulation. Literature suggests two alternative solutions to achieve a controlled condition. These options and the challenges are discussed as follows:
Sham stimulation: In a sham stimulation paradigm, the participant experiences a similar testing condition with no or minimal stimulation. A common way to achieve this is by tilting or rotating the stimulating coil away from the scalp. This results in same sound intensity and tactile sensation, but no current being actually delivered to the brain. This method does cause unnecessary stimulation of the brain. Some commercial sham coils do not need any coil maneuvering.
Another method to achieve sham stimulation involves disconnecting the coil from the stimulator while maintaining its position over the head. A second coil is connected to another stimulator with the same stimulation parameters but is discharged away from the participant (for example under a pillow). This allows the participant to experience the same coil position over the scalp and hear the clicking noise arising from the second coil without active current stimulation. However, this method does not create the fine tapping tactile sensation over scalp that one experiences with active stimulation. To counter this issue, Rossini et al. (2007) developed a real electromagnetic placebo device (REMP). It consists of a 3 cm thick compact wood attached to the real coil (Magstim figure-of-8 coil) with velcro straps that acts as a physical barrier to the magnetic field. The straps hoist a pair of electrodes connected to a stimulator that is used to pass a small electrical current synchronized with the TMS pulse. Though REMP is close to an ideal sham, this method could still induce a bias arising from the differential weights between sham (heavier due to added compact wood) and the figure-of-8 coil as well as the appearance of the sham coil. Borckardt (2011) designed a simple, portable inexpensive sham coil that conducts an electric current through electrodes attached to the scalp that are in sync with sham TMS pulses. The intensity of the sham coil is titrated to participants' sensation similar to that of the active stimulation.
Off-target stimulation: In off-target stimulation technique, the stimulation site is shifted to a brain region that is not directly involved in the task performance, called the “Empty Quarter.” Most studies have used vertex as the off-target stimulation site. Such a control site provides an identical sensory experience to that of the active stimulation; however, it is important that the control site is not directly or indirectly involved in the task under study. Moreover, this method, like that in the sham stimulation, exposes the participant to additional stimulation. For application of TMS in gait research, very few off-target stimulation sites are available as several regions of brain are involved during walking.,, In the next section, the factors involved in the application of repetitive TMS (rTMS) specifically during the walking task will be presented. The rTMS, as term suggests is the application of repetitive pulse in quick succession and further divided into high frequency (>1 Hz) and low-frequency rTMS (≤1 Hz).
On-line rTMS application: In this method of application, a fixed number of pulses are applied during the walking task. Walking is a horizontal translation of human body in space, but also causes a relatively small vertical displacement. The vertical displacement during walking compromises the consistency of the TMS application. Furthermore, the stepping pattern could be synchronized to external rhythmic clicking from the rTMS, resulting in walking that is “not free.”,, Moreover, on-line TMS application could draw attention on the walking task that typically requires little or no attention.
Off-line or event related application: In this method, a series of stimulation is applied prior to task performance and the neural disturbance should outlast the stimulation period. The spatial resolution of TMS is around 1 to 2 cm which means that the effect of induced current during rTMS could spread to the surrounding brain areas. This has been demonstrated during rTMS application over the hand motor cortex, but not yet reported in rTMS application to the leg motor cortex. Although not demonstrated in an experiment, we can hypothesize that the spread of stimulation is likely to occur when targeting leg motor cortex due to the proximity of the regions across the mid-sagittal fissure. To mitigate this issue and preferentially stimulate unilateral leg motor cortex, one can record muscle potential bilaterally and identify any spread of stimulation. Lastly, the variability in the application of rTMS could arise from the rapid fatigue induced in the arm of the operator when using a handheld coil. This alone or combined with patient's head movement may shift the focus of stimulation. Challenges with stimulation of the leg motor cortex will be discussed further down.
Challenges with Motor Threshold estimation: International Federation of Clinical Neurophysiology (IFCN, 1994) introduced motor threshold (MT) as a tool for studying cortical excitability., The MT is used to calculate dosage for therapeutic application, standardize stimulus intensities between individuals, and determine the intensity of conditioning stimuli in the paired-pulse application method. Mills (1997) used the term “cortico motor threshold” (CMT) as it stimulates both the cortical and spinal motor neurons which can independently interact and influence excitability threshold. Literature reports several methods of estimating the MT threshold.
The following section presents some of the most commonly deployed MT estimation methods and their drawbacks.
Motor threshold is defined as the lowest stimulus intensity (as a percentage of maximal stimulator output-MSO) required to induce MEPs in 5 out of 10 trials with at least 50 μV peak-to-peak amplitude in a relaxed target muscle called resting MT (rMT) or 200–300 μV amplitude during slight tonic contraction (20% of maximal strength) in an active target muscle called active MT (aMT). IFCN recommends beginning at the sub-threshold intensity of maximal stimulator output (%MSO) and gradually increase in increments of 5% until it evokes an amplitude of 50 μV than lower in steps of 1% till less than 5 out of 10 positive responses are recorded. The stimulus intensity plus 1 is defined as CMT. The drawback with the IFCN guidelines is the long time needed to determine CMT resulting in unnecessary stimulation. Rothwell (1999) suggested starting at the supramaximal intensity and decrease at an increment of 2% to 5% till a level is reached below which the reliable responses disappear. There are several drawbacks to Rossini–Rothwell algorithm. First, there is a lack of standard procedure to obtain supramaximal threshold. Second, the method is unclear on how to quantity of “slight” muscle contraction. Third, Rossini's method is based on a probability approach to estimate MT as opposed to a deterministic approach based on using a fixed number of pulses. Lastly, the method does not consider the stimulations has excitatory effect on the neurons but fails to meet the cut-off threshold (200 μV amplitude) to qualify for MT. Mills–Nithi proposed a more robust technique to separate MT at upper and lower level of stimulation intensity. The lower level MT is defined as the highest stimulus intensity with no motor response in 10 trials and upper MT as the lowest intensity with all 10 motor responses in 10 trials. An estimate of MT is obtained as an arithmetic mean of these two levels. Awiszuz method is based on the maximum likelihood strategy. In this method, the subsequent stimulus intensity is based on the response evoked by the preceding stimulus such that the MT is obtained after certain number of stimuli, known as the threshold hunting method., Parameter estimation by sequential testing (PEST) is based on the assumption that the probability of observing a motor response increases monotonically with increasing stimulus intensity and follows a sigmoid probability function. However, this model does not define the start of stimulus intensity to work around. Both the Awiszuz and PEST methods require long time to determine the motor threshold. Some studies have defined MT as the intensity at which visible muscle contraction is elicited 50% of the time in a set of stimuli. A study comparing the Rossini–Rothwell, Mills–Nithi, and visible contraction methods showed no difference in MT values. However, Rossini–Rothwell technique of adaptive staircase was recommended as it utilizes a comparatively lower number of stimuli.
The next section will present the technical challenges with the application of TMS during gait research.
Challenges due to devices: Coil positioning (orientation, tilting) is one of the major sources of error in MT measurements., A subtle deviation in position or orientation of the coil over the scalp may alter the induced electric field. Each muscle has an optimal site of stimulation in the cortex, known as “hot spot,” which is defined as an MEP with the greatest amplitude and shortest latency. Any shift from the “hot spot” decreases the likelihood of eliciting an optimal MEP. As the characteristic of an MEP determines MT, it is critical to maintain a stable coil to obtain an accurate response. However, the coil stability is threatened with long duration of stimulation and weight of the coil especially with handheld application.
The challenge with using MT when expressed as a percentage of MSO is that MSO is linked to the stimulator parameters such as stimulus waveforms, duration of pulse, and the coil properties (e.g., size, shape, number of copper windings). It is therefore difficult to compare across studies that use different stimulators or coil designs. To account for this variability, it is suggested that MT should be reported as a percentage of the maximum output voltage for that stimulator.
Accessibility and organization of the leg motor cortex: The cortical representation for the upper limbs is larger and therefore easily accessible compared to relatively small representation for the lower limbs. In addition, for a relative difference in the cortical representation, there is evidence to suggest that the neural control of upper limb is different from that of the lower limbs primarily involved in walking., TMS studies commonly target tibialis anterior muscle, due to its lower threshold for stimulation, larger MEP amplitude, wider representation compared to other leg muscles. fMRI studies have demonstrated variability in the depth within the interhemispheric sulcus and that the tibialis anterior recruitment activity can be recorded as far as the cingulate gyrus and over the convexity of the brain. This is also supported by intraoperative electrical stimulation that shows large, scattered area of tibialis muscle representation measuring around 15 to 34 mm below the cortical surface. Literature reports cerebral cortex to be 1 to 2 cm below the cortical bone and the leg motor area to be 3 to 4 cm from the scalp., The deeper location of leg motor cortex necessitates higher stimulation intensity compared to that needed for upper limb motor cortex.
It is further important to selectively stimulate unilateral leg motor area due to the close proximity regions between the hemispheres. Within the motor homunculus, muscles are organized in somatotopic fashion. Neuronal organization within leg motor cortex shows innervation by both crossed fibers from the contralateral hemisphere and uncrossed fibers from ipsilateral hemisphere. In addition, neurons in the leg motor cortex show a directional change with the majority of fibers aligned along medio-lateral than anterio-posterior orientation in the inter-hemispheric fissure., As the fibers travel down, the position and direction become angulated over the lateral ventricles from pyramidal decussation via internal capsule to spinal cord. This cytoarchitectural orientation is further complicated by the presence of sulci and gyri in the brain. The location and orientation of neurons in the leg motor cortex is complex and variable compared to relative perpendicular orientation of neurons in the hand motor cortex.,, The peripheral nerve studies suggest that optimal stimulation occurs when the current field is oriented along the direction of the nerve fiber.,,, For instance, there is preferential stimulation of right leg motor area using left to right current flow with a figure-of-8 coil. The tangential electric fields preferentially stimulate horizontal oriented neurons. The stimulation occurs at the point where the nerve fiber exits the field. Therefore, relative perpendicular orientation of neurons in the upper limb motor cortex requires less stimulus intensity as these fibred exit sharply out of the electrical field. The complex alignment of leg motor neurons affects the spatial distribution of the induced current. In addition, the long leg motor neurons pathway increases the probability of temporal dispersion of the corticospinal volley of signals and may therefore need greater stimulation intensity.
Technical challenges: Coil to cortex distance includes coil to scalp, scalp to cortex distance, and thickness of the skull. Scalp to cortex distance varies between subjects, across ages, and over different brain regions. For instance, scalp to cortex distance is greater in older adults due to cortical atrophy or children due to unmyelinated long fibers. Scalp thickness varies among different races. This variability in scalp to cortex distance will affect the induced secondary electric current. In addition, this change in distance may shift of stimulation site of pyramidal cells from transsynaptic activation to axon hillock thereby causing a change in the MEP latency. To account variable coil to scalp variability, it is recommended to increase intensity by 2.5% to 3% (2.9% or ~0.064 T) for every millimeter increase in distance between the coil and scalp. However, this guideline assumes that the rate of decay in the induced electric field between coil to scalp (air) and scalp to cortex and the intervening nervous tissue is identical. The decay gradient is 2.1% per mm for scalp to cortex distance and 2.8% per mm for scalp coil distance. Another study reports these values to be 1.5% per mm for scalp to cortex and 2.5% per mm for scalp to coil distance.
The other challenge with stimulating deeper brain areas is to minimize stimulation of the superficial structure. The H-coil is shown to effectively reach at a depth of 5 to 6 cm in healthy volunteers compared to 1.5 cm with a standard figure-of-8 coil with same level of superficial stimulation.,,
Coil positioning during gait task: One important challenge with targeting the leg muscles using TMS is to distinguish between muscle groups that are active at during various phases of walking. TMS application during gait requires stable positioning of the coil over the participant's head. Various devices to stabilize the coil over the head during a walking task are available; Balgrist harness (Balgrist Tec, Zurich, Switzerland),, modified hockey helmet with a special frame that allows free access to the scalp and special frame to mount the coil, and halo-vest system. Halo-vest system consists of a cyclist's helmet modified to provide free access to scalp and screwed to a frame that fixes the head to the thorax. The stimulation coil was mounted to the frame with an adjustable clamp. The feasible halo-vest system is shown to alter free gait pattern as it limits head and trunk movement, prevents visual scanning of the environment, and alters neck proprioceptive, visual as well as vestibular inputs.
The other challenges in use to TMS that are not limited to studying gait are the influence of hormones in women and men, diurnal variations, genetic make-up for instance increased excitability of motor cortex observed in people with Val66Met polymorphism (gene encoding for brain-derived neurotropic factors) and other genetic mutations, interaction with drugs such as antiepileptic drug sodium valproate,, and age and variation in scalp to cortex distance as in child and older adults.
This review discusses several important technological and methodological sources of variation in TMS research. While some sources of variations are inherent to the TMS techniques irrespective of the task, other sources arise from TMS application to gait. This review will help researchers choose an appropriate method when using TMS in gait research. While TMS is important to establish the direct involvement of brain areas in a task, the extent to which these variations impact the causal link between brain and task is difficult to estimate.
Financial support and sponsorship
Conflicts of interest
There are no conflicts of interest.