Participants
Five participants [ages 20–25 (mean 22.8 ± 2.2 years); 3 male, 2 females; 4 right handed, 1 left handed] were included in the study. This study was approved by University of Minnesota’s Institutional Review Board and all participants gave written informed consent to participate. Participants were physically and neurologically healthy and had no history of neurological disorders. Participants were also screened for medications contraindicated for other forms of non-invasive neuromodulation [14].
Experimental procedures
The study consisted of two magnetic resonance imaging (MRI) scanning sessions on separate days. The first session included a T1 anatomical scan and a functional scan with the finger tapping task (see below) to identify M1 thumb, index and middle finger representations. The thumb representation was then used as the target for the application of tFUS for the second session. In the second session, participants performed the same finger tapping task during either tFUS or sham neuromodulation. The order of tFUS and sham conditions was counterbalanced across participants.
Finger tapping task
Participants performed a visually cued finger tapping task using either the thumb, index, and middle fingers with their self-reported dominant hand. Participants lay supine in the MRI with their dominant arm supported with foam to ensure a comfortable position to tap their fingers on their thigh while limiting proximal arm and shoulder movement. Visual cues indicating the timing for tapping were presented using Cogent (www.vislab.ucl.ac.uk/cogent.php) for Matlab (MathWorks, Natick, MA, USA) and delivered using a projector to a screen that participants could see while inside of the bore of the MRI machine. The visual cues displayed the text (‘thumb’, ‘index’, or ‘middle’) with white block letter on a black background in the center of the screen with a large font, indicating the finger to be tapped paced at 1 Hz. This task used a block design with a single finger to be tapped for the duration of a block at the 1 Hz pace. Each finger was tapped for three blocks for a total of nine 30 s blocks, with 30 s rest blocks separating each finger tapping block (Fig. 1a). The ordering for the finger to be tapped per block was pseudo-randomly generated for each MRI scan where no finger would be tapped for three contiguous blocks.
Prior to scanning, participants practiced the finger tapping task to familiarize themselves with the task demands. To standardize movement range, participants were instructed to follow the visual prompts by extending and flexing the cued finger at the proximal phalanx while limiting movement of other fingers. Participants performed this practice session with feedback from the study staff to ensure the task would be performed properly while inside the scanner. Ultrasonic waveforms were delivered every two repetition times (TR, 2750 ms) for a total of 6 stimulations per 30 s block (54 total stimulations per scan). The tFUS condition involved acoustically coupling the active face of the ultrasound transducer to the scalp at the pre-determined neuronavigation (see below) site. To achieve acoustic coupling to the head, the volunteer’s hair was parted to expose the scalp and ultrasound gel was used to keep the hair out of the way and ensure proper coupling with the tFUS transducer. The transducer was also prepped with ultrasound gel on the surface that met the head, and was then placed on the exposed scalp and held in place using a secure head band. The sham condition involved turning off the transducer so that it would not deliver stimulation. Participants reported no auditory or tactile sensation from either the tFUS or sham condition as has previously been reported in similar setups outside of the MRI environment [1, 9].
tFUS waveform and delivery
The ultrasound transducer was a custom made [15] 30 mm diameter 7T MRI compatible single element focused 500 kHz with a focal length of 30 mm. The waveform used was the same as previously described [1]. This waveform was generated using a two-channel 2-MHz function generator (BK Precision Instruments, CA, USA). Channel 1 was set to deliver tFUS at a pulse repetition frequency (PRF) at 1 kHz and channel 2 was set to drive the transducer at 500 kHz in burst mode while using channel 1 as the trigger for channel 2. Channel 2 was set to deliver 180 cycles per pulse, and channel 1 was set to deliver 500 pulses, resulting in a 500 ms duration (Fig. 1b). Channel 2 output was sent to a 100 W linear amplifier (2100L Electronics & Innovation Ltd, NY, USA), with the output of the amplifier sent to the custom made tFUS transducer while using a Mini-Circuits (New York City, NY) 50-ohm low pass filter (1.9 MHz cutoff frequency) between the amplifier and the transducer at the patch panel to reduce radio frequency noise [16] and an “L” matching network to match the impedance of the RF amplifier and the transducer consisting of an inductor and capacitor arranged in the low pass form to also suppress higher order harmonics in the driving source [17].
Quantitative acoustic field mapping
The acoustic intensity profile of the waveform was measured in an acoustic test tank filled with deionized, degassed, and filtered water (Precision Acoustics Ltd., Dorchester, Dorset, UK). A calibrated hydrophone (HNR-0500, Onda Corp., Sunnyvale, CA, USA) mounted on a motorized stage was used to measure the acoustic intensity profile from the ultrasound transducer in the acoustic test tank at a 0.5 mm spatial resolution. Intensity parameters were derived from measured values of pressure using the approximation of plane progressive acoustic radiation waves. The ultrasound transducer was positioned in the tank using opto-mechanical components (Edmund Optics Inc., Barrington, NJ and Thorlabs Inc., Newton, NJ). Acoustic field scans were performed in the free water of the tank. Measurements in the acoustic tank revealed an spatial peak pulse average intensity (Isppa) of 16.95 W/cm2 and a mechanical index (MI) of 0.97 from the ultrasonic neuromodulation waveform in water. The − 3 dB pressure field was 3.83 mm in the X axis, 3.98 mm in the Y axis and 33.6 mm in the Z axis (Fig. 2). We have previously modelled the acoustic field through human skulls overlying the motor cortex demonstrating the skull to reduce peak pressure produced by the transducer in free water by a factor of 6–7, and it can be expected for the targeted region of the brain to experience pressure to be reduced as such [18]. In addition, the brain tissue and skull do not alter the beam path significantly [18, 19] or result in appreciable heating of the skin or skull bone [19].
tFUS targeting
The target for tFUS was chosen based on the isolated thumb fMRI representations found in the first MRI session (Fig. 3b). The thumb BOLD representation was loaded into a stereotaxic neuronavigation system (BrainSight; Rogue Research Inc, Montreal, Quebec, CA), and targets were created to guide tFUS based on the strongest BOLD signals in M1 with an approximate depth of ~ 30 mm (based on the focal length of the transducer) from the scalp on a per subject basis (Fig. 3b).
Quantitative modelling of ultrasound wave propagation
To better quantify the intracranial pressure in primary motor cortex from tFUS, a computational model was run to visualize and evaluate the wave propagation of tFUS across an example skull. The model was run using a magnetic resonance (MR) imaging and computerized tomography (CT) dataset taken from the Visible Human Project® [20]. The transducer was placed on the scalp site overlying the hand knob of the primary motor cortex. Simulations were performed using the k-Wave MATLAB toolbox [21] and modelling parameters and methods are detailed in [18]. The modelled beam is overlaid on an individual subject MRI image to show the ultrasound beam location relative to the thumb functional activity (Fig. 3a) and also to show the lateral resolution of the modelled beam relative to fMRI finger activations (Fig. 3c).
MRI acquisition parameters
All MRI scans were performed at the University of Minnesota’s Center for Magnetic Resonance Research on a 7T Siemens MRI scanner (Siemens Medical Solutions, Erlangen, Germany) using a Nova Medical 1 × 32 head coil (Wilmington, MA, USA). The fMRI scans were acquired using a gradient echo, echo planar image pulse sequence with the following parameters: repetition time (TR) = 2750 ms, echo time (TE) = 22 ms, flip angle = 70, field of view (FOV) = 192 mm × 192 mm, number of slices = 108, voxel size = 1.05 × 1.05 × 1.05 mm3, integrated parallel imaging technique (iPAT) = 3. Additionally, T1 anatomical scans were performed with the following parameters: TR = 3000 ms, TE = 3.28 ms, flip angle = 6, FOV = 192 mm × 216 mm, number of slices = 256, voxel size = 1 × 1 × 1 mm3.
BOLD fMRI data analysis
The fMRI data was processed in Analysis of Functional NeuroImages (AFNI) [22]. The data had 3D motion correction, linear and quadratic trends removed, a Gaussian filter with full width half maximum of 3 mm applied, slice timing correction, and distortion correction applied. A general linear model analysis was utilized to generate a statistical parametric map with a reference function generated by convolving the hemodynamic response function with the task function. This process was performed for all subjects’ fMRI data to isolate the individual representations of the thumb, index, and middle fingers using a threshold of t = 5 (p = 1e−6 uncorrected). To measure volume changes, a region of interest (ROI) was drawn around the pre-central gyrus (M1) to the depth of the central sulcus. Activated voxels (t = 5; p = 1e−6) in this ROI were used to calculate the activation volume in M1 due to the finger movement being performed for both the tFUS and sham condition. To test for differences between tFUS and sham neuromodulation, the total number of voxels that met this threshold within this ROI was subjected to a paired student’s t test.
For percent signal change analysis, we concentrated on a brain volume at the measured focal volume of the ultrasound beam (see Fig. 3). These coordinates were found for each subject and an ROI of 125 mm3 (5 × 5 × 5 mm) was drawn to encompass partial volume of the ultrasound pressure field. Based upon free water field ultrasound beam measurements, the FWHM volume of the beam was ~ 230 mm3. Percent signal change between tFUS and sham conditions were compared with a paired t test (N = 5). To further investigate the spatial selectivity of the tFUS effect, a 5 × 5 × 5 mm ROI was also placed at the region of strongest M1 activations for the index and middle finger representations in each participant to examine if tFUS has effects on these representations despite not being directly targeted for stimulation. Similar group (N = 5) paired t-tests were performed separately for the index and middle finger representations.
To test for potential downstream motor network effects as has previously been shown [11], we also examined the effect of tFUS to M1 on the SMA and ipsilateral PMd. The SMA and PMd were defined according to anatomical landmarks. Specifically, SMA included the volume between the precentral and central sulci down to the cingulate sulcus and laterally such that the ROI borders M1 and PMd. The PMd ROI included parts of the superior frontal gyrus and middle frontal gyrus lateral to the SMA and anterior to the pre-central sulcus. Data from the entire scanning session (9 on blocks; thumb, middle and index finger movement; 54 tFUS stimulations) was used in this analysis. We examined both volume and average percent signal from both the SMA and PMd volumes for each participant and each region was tested in a separate group (N = 5) paired t-test to assess differences between the tFUS and sham condition.