Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2025 Nov 1.
Published in final edited form as: Clin Neurophysiol. 2024 Sep 19;167:267–281. doi: 10.1016/j.clinph.2024.09.006

Low-intensity focused ultrasound to the insula differentially modulates the heartbeat-evoked potential: a proof-of-concept study

Andrew Strohman 1,9,10, Gabriel Isaac 1,4, Brighton Payne 1, Charles Verdonk 5,6,7, Sahib S Khalsa 5,8, Wynn Legon 1,2,3,4,9,10,11
PMCID: PMC11791892  NIHMSID: NIHMS2024333  PMID: 39366795

Abstract

Objective.

The heartbeat evoked potential (HEP) is a brain response time-locked to the heartbeat and a potential marker of interoceptive processing that may be generated in the insula and dorsal anterior cingulate cortex (dACC). Low-intensity focused ultrasound (LIFU) can selectively modulate sub-regions of the insula and dACC to better understand their contributions to the HEP.

Methods.

Healthy participants (n=16) received stereotaxically targeted LIFU to the anterior insula (AI), posterior insula (PI), dACC, or Sham at rest during continuous electroencephalography (EEG) and electrocardiography (ECG) recording on separate days. Primary outcome was HEP amplitudes. Relationships between LIFU pressure and HEP changes and effects of LIFU on heart rate and heart rate variability (HRV) were also explored.

Results.

Relative to sham, LIFU to the PI, but not AI or dACC, decreased HEP amplitudes; PI effects were partially explained by increased LIFU pressure. LIFU did not affect heart rate or HRV.

Conclusions.

These results demonstrate the ability to modulate HEP amplitudes via non-invasive targeting of key interoceptive brain regions.

Significance.

Our findings have implications for the causal role of these areas in bottom-up heart-brain communication that could guide future work investigating the HEP as a marker of interoceptive processing in healthy and clinical populations.

Keywords: focused ultrasound, insula, interoception, heartbeat-evoked potential, mental health, cingulate

1. INTRODUCTION

Interoception refers to an internal sense of the physiologic conditions of the body and relies on the transmission of neural signals between the brain and internal organs like the heart, lungs, gut, and vasculature (Berntson and Khalsa, 2021; Bonaz et al., 2021; Craig, 2002; Khalsa et al., 2018). Dysfunction in interoceptive processing has been broadly implicated in mental health and chronic pain disorders (Bonaz et al., 2021; Di Lernia et al., 2016; Khalsa et al., 2018), making efforts to modulate interoceptive signaling potentially promising for future therapeutic interventions.

Afferent information from internal organs travel via the vagus and glossopharyngeal nerves to the brainstem and thalamic targets and integrate with lamina I spinothalamic projections before relaying to the insula and dorsal anterior cingulate cortex (dACC) (Craig, 2009, 2002; Dum et al., 2009; Gasparini et al., 2020; Palma and Benarroch, 2014). Both the PI and the dACC receive direct afferent interoceptive information through two distinct thalamic nuclei while the AI communicates with PI and dACC without directly receiving this information, suggesting complementary but specific roles in interoceptive processing (Craig, 2004, 2002, 2014; Dum et al., 2009; Mandonnet et al., 2024). This makes insular subregions and the dACC intriguing targets for both pathophysiologic investigation and neuromodulation therapies across a range of mental health and chronic pain conditions that may involve brain-body dysregulation (Barrett and Simmons, 2015; Beissner et al., 2013; Seeley, 2019; Seth, 2013; Siddiqi et al., 2021; Taylor et al., 2023; Wager et al., 2013a).

The heartbeat-evoked potential (HEP) is a brain-derived electrophysiologic signal time-locked to each heartbeat that may indicate the central representation of signaling from the heart and baroreceptors (Coll et al., 2021; Engelen et al., 2023; Park and Blanke, 2019; Schandry et al., 1986). Alterations in the HEP have been implicated in chronic pain (Solcà et al., 2020) and mental health disorders like anxiety and depression (Coll et al., 2021; Pang et al., 2019; Terhaar et al., 2012; Verdonk et al., 2024), making the HEP a potential therapeutic biomarker across a wide range of clinical conditions. Intracranial recordings have also established the anterior insula (AI), posterior insula (PI), and dACC as either generating or contributing to the HEP (Mazzola et al., 2023; Park et al., 2018; Park and Blanke, 2019; Quadt et al., 2018), but it is presently unknown whether non-invasive focal neuromodulation can impact the HEP.

Low-intensity focused ultrasound (LIFU) is an increasingly investigated non-invasive neuromodulatory approach that leverages mechanical energy to transiently alter brain activity at a millimeter-sized focus with an adjustable depth (Darmani et al., 2022; Tyler et al., 2018). LIFU has been demonstrated to affect a range of physiologic markers when targeting both superficial and deep structures in humans that can also impact behavior (Legon et al., 2018a, 2014; Mueller et al., 2014; Nakajima et al., 2022; Sarica et al., 2022; Yaakub et al., 2023; Zeng et al., 2022). More specifically, recent work from our lab demonstrates LIFU can specifically target the PI, AI, and dACC with differential effects on subjective report of pain, pain-evoked electrophysiologic potentials, and heart rate variability (In et al., 2024; Legon et al., 2024; Strohman et al., 2024).

The purpose of this proof-of-concept study was to investigate site-specific roles of the PI, AI and dACC on the HEP, heart-rate and heart-rate variability (HRV), using LIFU to focally target these regions. We conducted this study during physiological resting conditions as many clinical brain stimulation therapies are applied at rest and evidence suggests dysfunction in neural and cardiac activity at rest across numerous clinical disorders (Kemp et al., 2014, 2010; Solcà et al., 2020; Zebhauser et al., 2023). We hypothesized LIFU would confer an inhibitory effect based on prior findings (Legon et al., 2024; Strohman et al., 2024), and specifically LIFU to the PI and dACC to attenuate the HEP amplitude as both regions receive cardiac afferents via the thalamus (Craig, 2004, 2002; Dum et al., 2009). By contrast, the AI does not receive direct cardiac input (Barrett and Simmons, 2015; Craig, 2002), thus we hypothesized that LIFU to the AI would not affect the HEP amplitude as compared to sham.

2. MATERIALS AND METHODS

2.1. Participants

The Institutional Review Board at Virginia Tech approved all experimental procedures (IRB #21–796). N=16 healthy participants (25.7 years ± 3.4 years; range (20–34); M/F 5/11), who met all inclusion/exclusion criteria provided written informed consent to all aspects of the study. Inclusion criteria were males and females ages 18–65 while exclusion criteria included contraindications to imaging (magnetic resonance imaging (MRI) and computed tomography (CT)), a history of neurologic disorder or head injury resulting in loss of consciousness for >10 minutes or drug dependence, and any active medical disorder or current treatment with potential central nervous system effects and no history of chronic pain.

2.2. Overall Study Design and Timeline

All participants completed a total of five visits in this sham-controlled cross-over design study. Following informed consent, the first visit comprised a structural brain MRI and CT for the purpose of LIFU targeting and acoustic modelling (see below). The remainder of the visits were formal testing sessions of LIFU to the anterior insula, posterior insula, dorsal anterior cingulate or Sham, randomized within and between subjects. A minimum of two days was required between visits to mitigate any potential carry-over effects.

This data was extracted from a larger study investigating LIFU modulation of pain processing (In et al., 2024). For the present study, we investigated the HEP during 10 minutes of LIFU application when participants were not completing any task. Figure 1 shows the LIFU conditions and LIFU delivery during the 10 min.

Figure 1. Study Design.

Figure 1.

Open in a new tab

A. Graphical depiction of formal testing days. Participants (N = 16) completed four conditions randomly assigned on four separate days: Low-intensity focused ultrasound (LIFU) to the anterior insula (AI), posterior insula (PI), dorsal anterior cingulate cortex (dACC) or Sham. Black dots on brains represent approximate locations of each target for the AI (left dorsal anterior short gyrus), PI (left dorsal posterior long gyrus), and dACC (Montreal Neurologic Institute (MNI) coordinates [0,18,30]) for all participants. Graphical brains were adapted from Biorender.com. B. For each session, data was collected over a 10 minute time window while participants sat at rest during continuous electroencephalogram (EEG) and electrocardiogram (ECG) recording. For each active day, a total of 100 LIFU stimulations were delivered over the 10-minute time window with a fixed interstimulus interval (ISI) of 5 seconds. Continuous auditory masking was delivered over the entire time window for all conditions. Blow-up of a single orange bar details the LIFU stimulation parameters. One second of LIFU was delivered with a fundamental frequency of 500 kHz, a pulse repetition frequency (PRF) of 1 kHz, a duty cycle (DC) of 36%, and a total burst duration (BD) of 1 second.

2.3. MRI and CT Acquisition

MRI data were acquired on a Siemens 3T Prisma scanner (Siemens Medical Solutions, Erlangen, Germany) at the Fralin Biomedical Research Institute’s Human Neuroimaging Laboratory. Anatomical scans were acquired using a T1-weighted MPRAGE sequence with a TR = 1400 ms, TI = 600 ms, TE = 2.66 ms, flip angle = 12 (degrees), voxel size = 0.5×0.5×1.0 mm, FoV read = 245 mm, FoV phase of 87.5%, 192 slices, ascending acquisition. CT scans were collected with a Kernel = Hr60 in the bone window, FoV = 250 mm, kilovolts (kV) = 120, rotation time = 1 second, delay = 2 seconds, pitch = 0.55, caudocranial image acquisition order, 1.0 mm image increments for a total of 121 images and scan time of 13.14 seconds.

2.4. Data Acquisition

2.4.1. Electroencephalography (EEG).

Surface EEG was collected using a DC amplifier (GES 400, Magstim EGI, Eugene, OR, USA) and Net Station TM 5.4 EEG software at a sampling rate of 1 kilohertz (kHz) with a 10 mm silver-silver chloride cup electrode placed on the vertex (Cz), referenced to the right mastoid. In this proof-of-concept study, we measured EEG signals using a single electrode placed at the Cz location. The Cz site was selected on the basis that 1) this location is one of the most commonly identified in meta-analytic studies of the HEP (Coll et al., 2021), and 2) Cz is a common site chosen in single electrode studies of visceral-evoked potentials (Aziz et al., 1995; Hobday et al., 2002; Hobson et al., 2004). The scalp was first prepped with a mild abrasive gel (NuPrep; Weaver and Company, Aurora, CO) and rubbing alcohol. The electrode was filled with a conductive paste (Ten20 Conductive; Weaver and Company, Auora, CO) and secured with medical tape. Electrode impedances were verified (<50 kΩ) before recording data, which was stored on a PC for offline analysis.

2.4.2. Electrocardiography (ECG).

Two ECG electrodes (MedGel™, MDSM611903) were placed on the anterior surface of both forearms immediately distal to the antecubital fossa (Einthoven lead I configuration) sampled at 1 kHz using a physiologic data acquisition box (Physio16, Magstim EGI, OR, USA) that has an integrated ground (GES 400, Magstim EGI, Eguene, OR, USA). Data was stored on a PC for offline analysis.

2.5. LIFU Transducers and Waveform

Two different transducers were used based upon individual participant head morphology and depth of target. For insula targets, a single-element 500 kHz transducer (Sonic Concepts H-281) was used with an active diameter of 45.0 mm and 38.0 mm focal length from the exit plane. The transducer also had a solid water coupling over the radiating surface to the exit plane. To target the dACC, a single-element 500 kHz transducer (Sonic Concepts H-104) was used with an active diameter of 64.0 mm and 52.0 mm focal length from the exit plane.

Focused ultrasound waveforms were generated using a dual channel function generator (BK 4078B Precision Instruments). Channel 1 was a 5Vp-p square wave burst of 1 kHz (N=1000) with a duty cycle of 36% used to gate channel 2 that was a 500 kHz sine wave. Channel 2 output was sent through a 100-W linear RF amplifier (E&I 2100L; Electronics & Innovation) before being sent to the transducers. LIFU was applied for 100 stimulations (burst duration = 1 sec) separated by a fixed inter-stimulus interval (ISI) of 5 seconds for a total application time of 10 minutes. These parameters were designed based on adapting concurrent LIFU parameters from prior work into a pre/post design for the larger study (Legon et al., 2014, 2018a, 2018b, 2024; Strohman et al., 2024). For both AI and PI, the applied external pressure was 380 kPa with a spatial peak-pulse average intensity (Isppa) of 4.2 W/cm2, a spatial peak temporal average (Ispta) of 1.51 W/cm2, and a mechanical index of (MI) of 0.2. For dACC, the applied external pressure was 400 kPa with an Isppa of 4.5 W/cm2, an Ispta of 1.62 W/cm2, and an MI of 0.23. These intensity values are not derated and were determined based on empirical measurements at the focus in free water using the acoustic test tank (see Empirical Acoustic Field Mapping section below for details). We also estimated LIFU pressure (kPa) at each brain target for each subject, which represent the derated pressures from skull and tissue attenuation.

Empirical Acoustic Field Mapping.

The acoustic intensity profile of the ultrasound 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 pressure from the ultrasound transducers in the acoustic test tank. The ultrasound transducers were positioned in the tank using a custom set-up and levelled to ensure the hydrophone was perpendicular to the surface of the transducer. XY and YZ planar scans were performed at an isotropic 0.25 × 0.25 mm resolution (Figure 2A & B). A voltage sweep from input voltages of 20–250 mVpp in 10mVpp increments was also performed to determine the necessary input voltage to obtain the desired extracranial pressure/intensity.

Figure 2. Transducer Characteristics and Acoustic Modelling.

Figure 2.

Open in a new tab

A. (Left) Pseudocolor XY and YZ empirical measurements showing the normalized pressure map for the transducer used for targeting anterior (AI) and posterior (PI) insula. (Right) Comparison of the YZ full-width half maximum (FWHM) beam profile of the transducer in free water (blue) compared to the YZ beam profile used for acoustic modelling. B. (Left) Pseudocolor XY and YZ empirical measurements of the normalized pressure map for the transducer used for targeting the dorsal anterior cingulate cortex (dACC). (Right) Comparison of the YZ beam profile of the transducer in free water (blue) compared to the YZ beam profile of the transducer used for acoustic modelling. C. Acoustic models using individual magnetic resonance and computed tomography scans from a representative subject showing 500 kHz LIFU targeting the AI. Color scale is pressure in kilopascals (kPa). Vertical dashed white lines in the coronal and transverse planes represent the slice taken for the sagittal view. D. Acoustic models using individual magnetic resonance and computed tomography scans from a representative subject showing 500 kHz LIFU targeting the PI. Color scale is pressure in kPa. Vertical dashed white lines in the coronal and transverse planes represent the slice taken for the sagittal view. E. Acoustic models using individual magnetic resonance and computed tomography scans from a representative subject showing 500 kHz LIFU targeting the dACC. Color scale is pressure in kPa. Vertical dashed white lines in the coronal and transverse planes represent the slice taken for the sagittal view.

2.6. LIFU Targeting and Application

Placement of the transducer on the scalp for each target site was aided using a neuronavigation system (BrainSight, Rogue Research, Montreal, QUE, CAN). This system was employed before each study visit to calibrate the LIFU transducer as well as the register the participant’s head to an uploaded structural MRI using high-definition infrared tracking. This permitted real-time tracking of the ultrasound transducer and participant’s head positions with high spatial precision. The free water focal length of the transducer from the exit plane was set in Brainsight for each transducer to allow the researcher to visualize the idealized focal spot during targeting. Targeting LIFU to each brain region through the skull was verified with acoustic modelling (details below). The transducer was coupled to the head using conventional ultrasound gel and our custom mineral oil/polymer coupling pucks (Strohman et al., 2023). These pucks have negligible attenuation at 500 kHz, These pucks are viscous gel polymers that have negligible attenuation at 500 kHz, can conform to shape of the scalp, and are made with varying stand-off heights that allow for precise axial (depth) targeting based on individual insular and dACC target depths from the scalp (Strohman et al., 2023) (Table A.1). For both AI, PI, and dACC conditions, an appropriate coupling puck was made so the focal spot of the transducer was exactly overlaid on the insular or dACC target. Each participant’s left AI (dorsal anterior short gyrus) and PI (dorsal posterior long gyrus) target was identified with the aid of an insular atlas [46] and depth was measured from the scalp. For dACC, MNI coordinate [0,18,30] was used based upon Neurosynth reverse inference software for the term ‘pain’ and prior literature on the fMRI signature of heat pain (Wager et al., 2013b; Yarkoni et al., 2011). LIFU was only delivered if placement error on the scalp was < 3 mm.

2.7. Auditory Masking

Participants were given headphones connected to a tablet that had a multi-tone white noise generator app. There were a variety of sound options that could be mixed and layered on top of each other to create a multitone that has been established to effectively mask a 1 kHz pulsed ultrasound soundwave (Liang et al., 2023). The volume was set to a level where participants could not hear normal conversations with the intensities ranging from 70 to 75 dB.

2.8. Questionnaires

2.8.1. Review of Symptoms

A review of symptoms questionnaire (Legon et al., 2020) asked about the presence of various symptoms and their severity (absent / mild / moderate / severe) scored on a scale of 0–3. This questionnaire was collected at the beginning of each visit and 30 minutes after the LIFU or Sham application in order to indicate any changes of symptoms from the intervention. Counts for each symptom at each severity level were calculated for both pre and post LIFU for each condition (AI, PI, dACC, Sham). Counts are also calculated adjusting for the presence of symptoms after LIFU compared to symptoms present before LIFU to better understand if new symptoms occurred or symptoms reduced after the intervention to each target site (Figure A.2).

2.8.2. Auditory Masking

An auditory masking questionnaire (Legon et al., 2024; Strohman et al., 2024) was administered at the end of each visit (sessions 2–5) with the following questions: “I could hear the LIFU stimulation”, “I could feel the LIFU stimulation”, and “I believe I experienced LIFU stimulation.” Participants were asked to rate each question on a 7-point Likert scale: strongly disagree / disagree / somewhat disagree / neutral / somewhat agree / agree / strongly agree. To statistically test for differences between conditions, scores for each question (“Hear”, “Feel”, “Believe”) were converted to numerical values from 0–6 and compared across conditions using a kruskal-wallis test at a significance level of p < 0.05. Data are reported as the mean ± SEM for each condition.

2.9. Formal Testing Days

On each formal testing day, participants were seated in a comfortable chair and connected to continuous EEG and ECG (see above). After the initial set of tasks prior to LIFU (see In et al., 2024), LIFU was stereotaxically targeted to either the AI, PI, dACC or Sham. Participants were also given headphones to play the masking auditory stimulus throughout the entire LIFU application period (Figure 1A). As outlined above, LIFU was delivered with a burst duration of 1 second and an ISI of 5 seconds for a total of 100 stimulations for a total duration of 10 minutes (Figure 1A&B). The beginning and end of LIFU stimulation denoted the EEG and ECG data of interest for pre-processing and analysis.

Sham Condition

The Sham condition was an inactive sham, where all procedures were identical to real visits (including the use of neuronavigation, the selection of the proper gel puck, and the application of continuous auditory masking), except no LIFU was actually delivered. For the Sham visit, the transducer was randomly placed at one of the active sites which was randomized between participants. The auditory masking questionnaire (AMQ) was used after each session to query on successful masking.

2.10. Data Pre-Processing

2.10.1. Electrocardiography

ECG data was first filtered from 0.3–35 Hz using a 3rd order Butterworth filter and movement artifacts were manually removed. R-R peaks were then extracted using the automated detection program findpeaks in Matlab and confirmed manually. Only R-R peaks for the time window during LIFU application were used for data analysis.

2.10.2. Electroencephalography

EEG pre-processing was adapted from Petzschner et al. (2019) (Petzschner et al., 2019). EEG data were preprocessed using custom scripts written in Matlab 2022a® (The MathWorks, Inc., Natick, MA). Data were band-pass filtered from 0.3–35 Hz using a third-order Butterworth filter and the filtfilt function. The 0.3 Hz lower bound for the bandpass filter was chosen based on previous literature suggesting higher high-pass values can introduce artifact into the signal (Petzschner et al., 2019; Tanner et al., 2015; Widmann et al., 2015). EEG data were epoched around each R-peak from 100 ms before to 667ms after the R-peak. 667 ms was chosen as this was the maximum value that still included individuals with a resting heart-rate < 90 bpm. Trials were rejected for either of two reasons: 1) If greater than two percent of the data across the epoch’s time window exceeded three standard deviations of the mean HEP amplitude or 2) If the instantaneous heart rate at any time exceeded 90 beats per minute (bpm) for that trial. In order to avoid artifact from preceding heartbeats, we did not baseline correct our epochs as previously performed (Petzschner et al., 2019). Based on the above criteria, the median (interquartile range (IQR)) number of trials removed was 52.5 (35), 56.5 (18), 57 (49.5), and 54 (22) for AI, PI, dACC, and Sham conditions, respectively (range: 7–416). A Kruskal-Wallis test revealed no significant difference between conditions (Chi-square = 0.27(3), p = 0.98). The total number of trials removed accounted for 12.16% of the data across all participants and conditions. This is a similar percentage of removed trials compared to prior literature (Coll et al., 2021).

The mean ± SD final number of trials included in the analysis was 539.6 ± 85.9, 500.5 ± 69.1, 510.8 ± 70.1, and 529.9 ± 73.1 for AI, PI, dACC, and Sham conditions, respectively. A one-way ANOVA revealed no significant difference between conditions (F = 1.70 (3,45), p-value = 0.18). The time window of interest (TOI) for HEP quantification was 200 to 667ms after the R-peak. This was based on Coll et al. (2021) as 200ms post R-peak is the earliest time HEP effects have been reported, and which corresponds to the time of increased baroreceptor firing when the systolic blood outflow stretches the walls of the aortic arch and carotid sinus (Coll et al., 2021; Gray et al., 2010). In addition, the influence of cardiac field artifact (CFA) on HEP amplitude is presumed to be limited from 200ms post R-peak (Coll et al., 2021; Kern et al., 2013; Petzschner et al., 2019). The final time point follows the approximate last time point where HEP effects have been previously reported and is based on a maximum resting heart rate of 90 bpm (Petzschner et al., 2019).

2.1. Data Analysis

2.11.1. Acoustic Modelling

Computational models were developed using individual subject MR and CT images to evaluate the wave propagation of LIFU across the skull and the resultant intracranial acoustic pressure maps. Simulations were performed using the k-Wave MATLAB toolbox (Treeby and Cox, 2010), which uses a pseudospectral time domain method to solve discretized wave equations on a spatial grid. CT images were used to construct the acoustic model of the skull, while MR images were used to target LIFU at either the AI, PI, or dACC target, based on individual brain anatomy. Details of the modeling parameters can be found in Legon et al. (2018) (Legon et al., 2018a). CT and MR images were first co-registered and then up-sampled for acoustic simulations and the acoustic parameters for simulation were calculated from the CT images. The skull was extracted manually using a threshold intensity value and the intracranial space was assumed to be homogenous as ultrasound reflections between soft tissues are small (Mueller et al., 2016). Acoustic parameters were calculated from CT data assuming a linear relationship between skull porosity and the acoustic parameters (Aubry et al., 2003; Marquet et al., 2013). The computational model of the ultrasound transducer used in simulations was constructed to recreate empirical acoustic pressure maps of focused ultrasound transmitted in the acoustic test tank similar to previous work (Legon et al., 2018a, 2018b, 2024; Mueller et al., 2017). For each subject and each stimulation site, the estimated LIFU pressure at the beam focus in kilopascals (kPa) was extracted and used for later analysis.

2.11.2. HEP Time Domain Analysis

To characterize LIFU-induced changes in HEP amplitude, we utilized non-parametric permutation testing with custom Matlab™ scripts following procedures recommended by Maris and Oostenveld (2007) (Maris and Oostenveld, 2007). We used the F-statistic from a one-way ANOVA with the four conditions (AI,PI, dACC, Sham) across the epoched EEG time window (100ms before to 667ms post R-peak), with 1000 permutations at each millisecond at a significance level of p< 0.05. Data is displayed without any cluster threshold in order to visualize any significant time points. The effect size for the average HEP amplitude across each significant time window is reported as the eta-squared (ŋ2). Time points before 200ms post R-peak (despite being in the window of the CFA) were included in the analysis to identify any time points with significant differences within the CFA. Any significant findings were used for control analyses (see below) to test whether these differences explain observed differences in the HEP (200 – 667 ms). Post-hoc comparisons on the average amplitude for each time window with significant differences (p < 0.05) in the TOI of the HEP were explored using Tukey’s test.

2.11.3. Control Analyses

Control analyses were adopted from Petzschner et al. (2019), with a couple of additions (Petzschner et al., 2019). We wanted to rule out that changes in HEP amplitude were not caused by changes in cardiac electrical activity, as measured by the ECG, which may propagate throughout the body and affect EEG scalp potentials. For this purpose, we epoched the ECG data using the same approach as for the EEG data (see above), and we tested the condition effect (AI vs. PI vs. dACC vs. Sham) on the averaged ECG waveform. To this end, the F-statistic from a one-way ANOVA with the four conditions (AI, PI, dACC, Sham) was tested across the entire window (−100ms to +667ms around the R-peak) using 1000 permutations at each millisecond at a significance level of p < 0.05. Data is displayed without any cluster threshold in order to visualize any significant time point differences in the ECG.

As an additional check to rule out potential confounding cardiac effects, we extracted ECG amplitudes from the same time window as any observed significant findings in the HEP data. We then employed linear regression on the mean value from the time windows, testing whether the change in amplitude in the HEP data was explained by any change in amplitude in the ECG data at these specific time points. Data is reported as the adjusted R2, F-statistic, and p-value at a significance level of p < 0.05.

For any significant findings in the EEG data during the CFA, the same approach was used as above with the ECG data. We averaged across all significant time points within the significant CFA time window, subtracted across the conditions identified from post-hoc testing in the EEG data at each significant time window in the TOI of the HEP, and employed linear regression to test whether the change in amplitude in the HEP window was explained by any change in amplitude in the CFA window. Data is reported as the adjusted R2, F-statistic, and p-value at a significance level of p < 0.05.

2.11.4. Analysis of Pressure.

Estimated LIFU pressures for each subject at each target were tested across conditions to evaluate for significant differences using a one-way ANOVA with factor CONDITION (AI, PI, and dACC) at a significance level of 0.05. With the presence of significant differences, we employed an analysis of co-variance (ANCOVA) model for each significant time window in the TOI of the HEP to account for pressure differences between conditions. The outcome variable was average HEP amplitude differences relative to sham. The predictor variables were condition (AI, PI, and dACC), pressure in kilopascals, and the condition*pressure interaction. The outcome of each model is reported with the F-statistic (DF) and p-value at a significance level of 0.05 for each variable and the interaction term. Tukey-Kramer post hoc tests were employed for any significant findings.

2.11.5. Analysis of Heart Rate & Heart Rate Variability.

We further explored whether LIFU had any effects on heart rate and heart rate variability (HRV), defined as the change in time intervals between consecutive heartbeats (Shaffer and Ginsberg, 2017). We first calculated median heart rate to investigate if there were any differences between conditions. We used a one-way kruskal-wallis test with main factor CONDITION (AI, PI, dACC, Sham). Data is reported as the median (IQR) with the chi-square (df), p-value at a significance level of p < 0.05, and the eta-squared (ŋ2). We also calculated four HRV metrics during the period of LIFU application: the standard deviation in normal sinus beats (SDNN), low frequency (LF) power, high frequency (HF) power, and the LF/HF ratio (Shaffer and Ginsberg, 2017). The SDNN is a time-domain metric and was calculated as the standard deviation of the time in milliseconds between consecutive normal sinus beats. For frequency domain metrics (LF power, HF power, LF/HF ratio), R-R interval data during LIFU or sham application was cubic spline interpolated and transformed using the fast-fourier transform. The LF power (0.04–0.15 Hz) and HF power (0.15–0.4 Hz) bands during LIFU or Sham application were calculated for each subject for each condition (Shaffer and Ginsberg, 2017). The LF/HF ratio was calculated by dividing the LF by the HF power. All HRV metrics were tested using a one-way ANOVA with main factor CONDITION (AI, PI, dACC, Sham). Data is reported as the mean ± SEM, the F-statistic (df) at a significance level of p < 0.05, and the eta-squared (ŋ2).

3. RESULTS

3.1. Insula & dACC target depths

The mean ± SD depth from the scalp to each target was 34.5 ± 2.6 mm, 39.4 ± 2.6 mm, and 47.3 ± 3.4 mm for the AI, PI, and dACC targets, respectively (Table A.1). Individual participant depths and means for each sex can be seen in Table A.1.

3.2. Acoustic Modelling

For the insula transducer, the beam profile in free water had a lateral full-width at half maximum (FWHM) resolution in the lateral (XY) resolution ± 1.7 mm at the axial (Z) beam focus and a FWHM axial (YZ) resolution of ± 11.5 mm with a focal depth of 38 mm from the exit plane of the transducer (Figure 2A). For the dACC transducer, the beam profile in free water revealed a FWHM lateral (XY) resolution of ± 1.5 mm at the axial (Z) beam focus and a FWHM axial (YZ) resolution of ± 10 mm with a focal depth of 52 mm from the exit plane of the transducer (Figure 2B). The volumes these transducers overlaid well with the individual subject target depths (Table A.1).

The axial FWHM comparison between the empirical measurements in free water and the modelled waveform for both transducers demonstrated good agreement validating use in the acoustic models (Figures 2A&B). Acoustic models for each site (AI, PI, dACC) from representative subjects can be seen in Figures 2C–E.

3.3. HEP Time Domain Analysis

Permutation analysis (permutations = 1000, 1ms intervals, p < 0.05, no cluster threshold) revealed three distinct time windows with significant differences between conditions (Figure 3A & B). The first window within the TOI of the HEP (200 – 667 ms) was a 13-millisecond window from 274 to 286 milliseconds after the R-peak. Of the 13ms, 9 non-consecutive time-points were significant and the p-values for the entire 13ms window ranged from 0.038 to 0.057 (ŋ2 = 0.09) (Figure 3A & B). As such, we used the entire window for subsequent post-hoc testing and control analyses. Post-hoc testing revealed significant differences between PI and Sham (p = 0.033) (Figure 3C). The second significant window within the TOI of the HEP was a 14ms time window from 648 to 661ms post R-peak (all p-values < 0.05, ŋ2 = 0.11) (Figure 3A & B). Post-hoc testing revealed a significant difference between PI and dACC conditions (p=0.014), but not Sham (Figure 3C). A plot depicting F-statistics and p-values from permutation testing across time can be seen in Supplemental Figure A.2A.

Figure 3. Analysis of HEP amplitudes.

Figure 3.

Open in a new tab

A. Group (N=16) mean ± SEM heartbeat-evoked potential (HEP) amplitudes across time. X-axis is time in milliseconds (ms) and y-axis is amplitude in microvolts (μV). Grey background represents the period of the cardiac field artifact while the white represents the HEP time window of interest (TOI). Vertical black bars at the top represent significant time points from permutation analysis (1000 permutations, p < 0.05). Vertical colored bars represent significant differences across conditions from post-hoc testing at each HEP time window of interest. F-statistics and p-values from permutation testing can be seen in Supplemental Figure A.3A. (Below) Magnified Group (N=16) mean ± SEM heartbeat-evoked potential (HEP) traces around the time periods of first (left) and second (right) time periods of significant results. Black arrow points to the same significant time window in A. B. (Left) Group (N=16) mean ± SEM HEP amplitudes (μV) across conditions within the first significant period in the HEP time window of interest. (Right) Group (N=16) mean ± SEM HEP amplitudes (μV) across conditions within the second significant period in the HEP time window of interest.

3.4. Control Analyses

In order to test for significant differences in the ECG trace within the TOI of the HEP, we ran the same permutation analysis (permutations = 1000, 1ms intervals, p < 0.05, no cluster threshold) across groups in the ECG data. There were no significant differences across any of the conditions (Figure 4A). A plot depicting F-statistics and p-values from permutation testing across time can be seen in Supplemental Figure A.2B.

Figure 4. Control analyses on ECG and cardiac field artifact.

Figure 4.

Open in a new tab

A. Group (N=16) mean ± SEM electrocardiogram (ECG) amplitudes across time. X-axis is time in milliseconds (ms) and y-axis is amplitude in microvolts (μV). Grey background represents the equivalent period of the cardiac field artifact from EEG data while the white represents the equivalent heartbeat-evoked potential (HEP) time window of interest. F-statistics and p-values from permutation testing can be seen in Supplemental Figure A.3B. B. (Left) Linear regression comparing the average change in the HEP amplitude to the average change in the electrocardiogram (ECG) amplitude, both across the first significant HEP time window for posterior insula (PI) relative to Sham. X-axis is the change (PI-Sham) in the ECG amplitude (μV) and y-axis is the change (PI-Sham) in the HEP amplitude (μV). Solid red line is the fitted regression line and the dotted lines represent the 95% confidence bounds. P = p-value. (Right) Linear regression comparing the average change in HEP and ECG amplitudes across the second significant HEP time window for PI relative to dorsal anterior cingulate (dACC) conditions. C. (Left) Linear regression comparing the average change in the HEP amplitude across the first significant HEP time window to the average change in amplitude across the significant cardiac field artifact (CFA) window for PI relative to Sham conditions. X-axis is the change (PI-Sham) in the CFA amplitude (μV) and y-axis is the change (PI-Sham) in the HEP amplitude (μV). Solid red line is the fitted regression line and the dotted lines represent the 95% confidence interval. P = p-value. (Right) Linear regression comparing the average change in HEP amplitudes across the second significant HEP time window and to the average change in amplitude across the significant CFA time period for PI relative to dACC conditions.

As an additional check to ensure changes in the ECG amplitudes were not driving observed changes in the HEP, we employed linear regression to compare the observed differences in HEP to differences in ECG amplitudes averaged across each significant time window. For the first significant window within the TOI of the HEP, changes in ECG amplitudes (PI-Sham) did not predict changes in the HEP amplitudes (PI-Sham) (R2adj= 0.06, F(1,14) = 2.03, p = 0.18) (Figure 4B left). For the second significant window within the TOI of the HEP, ECG amplitudes (PI-dACC) did not predict changes in the HEP amplitudes (PI-dACC) (R2adj = −0.06, F(1,14) = 0.16, p = 0.69) (Figure 4B right).

In the permutation analysis (Figure 3A), there were four consecutive milliseconds within the CFA showing significant group differences from 33 to 36ms post R-peak (all p-values < 0.05). This time window was used for control analyses to compare against observed changes in the HEP to check if changes in the significant CFA window predicted changes in either significant window of the HEP. We used the PI-Sham and PI-dACC differences in the HEP and CFA amplitudes averaged across the first and second significant time window, respectively. For the first significant window within the TOI of the HEP, changes in CFA amplitudes (PI-Sham) did not predict changes in the HEP amplitudes (PI-Sham) (R2adj = −0.03, F(1,14) = 0.52, p = 0.48) (Figure 4C left). For the second significant window within the TOI of the HEP, changes in CFA amplitudes (PI-dACC) did not predict changes in the HEP amplitudes (PI-dACC) (R2adj = −0.01, F(1,14) = 0.82, p = 0.38) (Figure 4C right).

3.5. Modelling Effects of Pressure

The mean ± SEM estimated LIFU pressure at the brain target across conditions was 179.75 ± 20.45, 262.81 ± 16.13, and 115.06 ± 13.23 kilopascals (kPa) for AI, PI, and dACC conditions, respectively. A one-way ANOVA demonstrated a significant difference between conditions (F = 22.87 (2,30), p-value < 0.001). Post-hoc testing revealed the PI was significantly higher than both AI and dACC while the AI was significantly higher than the dACC (Figure 5A). To examine how pressure and condition explain observed differences in the HEP, we employed an ANCOVA with grouping factor condition, the continuous covariate pressure, and the condition*pressure interaction.

Figure 5. Modelling the effects of pressure on observed HEP differences.

Figure 5.

Open in a new tab

A. Group (N=16) mean ± SEM estimated LIFU pressure in the brain in kilopascals (kPa) across conditions. B. Visualization of the significant condition*pressure interaction identified from the analysis of covariance (ANCOVA) for the first significant heartbeat-evoked potential (HEP) time window. X-axis is pressure in kilopascals (kPa) and the y-axis is the change in the HEP amplitudes relative to sham in microvolts (μV). AI = anterior insula (red); PI = posterior insula (purple); dACC = dorsal anterior cingulate cortex (green). C. Visualization of the significant effect of condition identified from the ANCOVA for the second significant heartbeat-evoked potential (HEP) time window. Bars represent the Group (N=16) marginal mean ± SEM for each condition. Note only the marginal means from the ANCOVA model for the effect of condition are shown as there was no main effect of pressure and no condition*pressure interaction as in B. Y-axis is the change in the HEP amplitudes relative to sham (μV). AI = anterior insula (red); PI = posterior insula (purple); dACC = dorsal anterior cingulate cortex (green). Horizontal black bars and p-values represent results from Tukey-Kramer post hoc testing of the marginal means from the ANCOVA model.

For the first significant time window, the analysis of co-variance (ANCOVA) showed a significant condition*pressure interaction (F(2,42) = 3.52, p = 0.039), but no main effect of condition (F(2,42) = 0.23, p = 0.79) or pressure (F(1,42) = 3.12, p = 0.085). The mean ± SEM slopes were 0.0095 ± 0.0040, −0.0069 ± 0.0042, and −0.0026 ± 0.0047 for the AI, PI and dACC conditions, respectively. Tukey-Kramer post hoc testing revealed significant differences in the slope (ẞdiff (95% CI)) between AI and PI conditions (ẞdiff = 0.016 (0.00060, 0.032), p = 0.041), but not between AI and dACC (ẞdiff = 0.012 (−0.0059, 0.030), p = 0.25) or PI and dACC conditions (ẞdiff = −0.0043 (−0.024, 0.015), p = 0.85). Figure 5B depicts the significant condition*pressure interaction.

For the second time window, the ANCOVA showed a significant main effect of condition (F(2,42) = 3.32, p = 0.046) but no main effect of pressure (F(1,42) = 1.06, p = 0.31) or the condition*pressure interaction (F(2,42) = 1.79, p = 0.18). The mean ± SEM marginal means were −0.15 ± 0.17, −0.35 ± 0.26, and 0.68 ± 0.28 for the AI, PI and dACC conditions, respectively. Tukey-Kramer post hoc testing of the marginal means extracted from the model (mean difference (95% CI)) showed significant differences between PI and dACC conditions (1.02 (0.09, 1.97), p = 0.030), and AI and dACC conditions (0.83 (0.026, 1.63), p = 0.042), but not between AI and PI conditions (0.20 (−0.56, 0.96), p = 0.79). Figure 5C depicts the marginal mean HEP amplitudes accounting for pressure.

3.6. Heart Rate and Heart Rate Variability.

The median (IQR) heart rate during LIFU application was 68.59 (16.8), 68.17 (14.64), 71.34 (12.85), and 68.64 (13.62) for AI, PI, dACC, and Sham conditions, respectively. A kruskal-wallis test revealed no significant differences between conditions (chi-square = 0.94 (3,60), p-value = 0.81, ŋ2 = 0.01).

The mean ± SEM SDNN (ms) during LIFU was 61.88 ± 5.56, 78.64 ± 6.31, 69.15 ± 5.73, and 68.11 ± 5.78 for AI, PI, dACC, and Sham conditions, respectively. A one-way ANOVA revealed no significant differences between conditions (F(3,45) = 2.78, p = 0.052, ŋ2 = 0.07).

The mean ± SEM HF power (absolute units or a.u.) during LIFU was 0.0031 ± 0.0005, 0.0040 ± 0.0005, 0.0031 ± 0.0003, and 0.0035 ± 0.0004 for AI, PI, dACC, and Sham conditions, respectively. A one-way ANOVA revealed no significant differences between conditions (F(3,45) = 2.75, p = 0.054, ŋ2 = 0.05).

The mean ± SEM LF power (a.u.) during LIFU was 0.0048 ± 0.0005, 0.0058 ± 0.0006, 0.0053 ± 0.0006, and 0.0058 ± 0.0007 for AI, PI, dACC, and Sham conditions, respectively. A one-way ANOVA revealed no significant differences between conditions (F(3,45) = 1.31, p = 0.28, ŋ2 = 0.03).

The mean ± SEM LF/HF ratio during LIFU was 1.73 ± 0.16, 1.69 ± 0.23, 1.88 ± 0.25, and 1.81 ± 0.21 for AI, PI, dACC, and Sham conditions, respectively. A one-way ANOVA revealed no significant differences between conditions (F(3,45) = 0.32, p = 0.81, ŋ2 = 0.01).

3.8. Report of Symptoms

The raw symptoms reported before and after LIFU for each condition are presented in Figure 6A & B. The most common symptom before or after any intervention was sleepiness. Prior to LIFU, one subject for both the PI and dACC conditions reported severe sleepiness. This was not present after LIFU application. In addition, we adjusted the symptoms present after LIFU with the symptoms present before LIFU to get an indication of new symptoms following the intervention (Figure 6C). After adjustment, there were no new moderate or severe symptoms across any condition for all queried symptoms. The most common new symptom was sleepiness with 1, 3, 3, and 5 new mild reports for the AI, PI, dACC, and Sham conditions, respectively. The most commonly reduced symptom was also sleepiness with 4, 6, 5, and 2 subjects demonstrating a reduction in sleepiness for the AI, PI, dACC, and Sham conditions, respectively. A full breakdown of the counts for each symptom can be seen in Figure 6.

Figure 6. Report of symptoms questionnaire results.

Figure 6.

Open in a new tab

Counts for each queried symptom and severity level for each condition. From left to right, the columns represent the anterior insula, posterior insula, dorsal anterior cingulate, and Sham conditions. Severity levels of each symptom are represented as absent (blue), mild (green), moderate (yellow), and severe (red). A. Presence of symptoms prior to LIFU or Sham application. B. Presence of symptoms after LIFU or Sham application. C. Presence of symptoms after LIFU or Sham application adjusted for symptoms present prior to LIFU or Sham application. Positive values represent counts of new symptoms for each severity level and negative values represent counts of reduction in symptoms at each severity level. No change is not represented.

3.7. Auditory Masking

For the question “I could hear the LIFU stimulation,” the mean ± SEM score was 1.44 ± 0.48, 2.00 ± 0.55, 2.94 ± 0.56, and 1.94 ± 0.50 for AI, PI, dACC, and Sham conditions, respectively. A kruskal-wallis test revealed no significant difference between conditions (chi-square = 3.85 (3,60), p = 0.28) (Figure A.1A).

For the question “I could feel the LIFU stimulation,” the mean ± SEM score was 0.88 ± 0.30, 0.81 ± 0.28, 0.69 ± 0.28, and 0.56 ± 0.22 for AI, PI, dACC, and Sham conditions, respectively. A kruskal-wallis test revealed no significant difference between conditions (chi-square = 0.73 (3,60), p = 0.87) (Figure A.1B).

For the question “I believe I experienced LIFU stimulation,” the mean ± SEM score was 3.31 ± 0.41, 3.13 ± 0.33, 2.94 ± 0.40, and 2.13 ± 0.33 for AI, PI, dACC, and Sham conditions, respectively. A kruskal-wallis test revealed no significant difference between conditions (chi-square = 6.7 (3,60), p = 0.08). Since the results for this question were trending, posthoc analysis revealed the effect was driven by differences between AI and Sham which was not the active site that drove any of the above results (Figure A.1C).

4. DISCUSSION

In this proof-of-concept study, we investigated the effect of 500 kHz LIFU delivered to either the AI, PI, and dACC compared to Sham on the amplitude of the heartbeat evoked potential, heart rate and heart-rate variability. We identified two distinct time windows in the EEG HEP data with significant group differences that could not be explained by changes in the electrocardiogram or by changes in the cardiac field artifact. These included an early time window around 280ms post R-peak where LIFU to PI was lower than Sham and a later time window around 650ms post R-peak where LIFU to PI was lower than LIFU to dACC. LIFU to the AI and dACC had no effect on the HEP at any time point. We also found no effect of LIFU to any of the targets on mean heart rate or time and frequency domain metrics of heart-rate variability.

4.1. LIFU to the PI attenuates afferent cardiovascular information

We identified two distinct time windows in the EEG data around 280ms and 655ms after the R-peak where LIFU to the PI decreased the HEP amplitude relative to Sham and dACC conditions, respectively. These changes were not related to changes in the CFA or ECG data. Including pressure as a covariate in the first time window showed a condition*pressure interaction where increased estimated LIFU pressure at the intracranial target was associated with greater decreases in the HEP amplitude for the PI condition. There was no significant impact of pressure in the second time window, but the original difference between PI and dACC was conserved. The HEP is commonly considered to reflect cortical processing of afferent cardiac or baroreceptor activity (Coll et al., 2021; Park and Blanke, 2019). Afferent cardiovascular information the heart and baroreceptors travels to the nucleus of the solitary tract (NTS) where it integrates information with signals from the Lamina I spinothalamic system in the ventrolateral medulla, parabrachial nucleus, periaqueductal gray, and thalamus, before relaying to the insula and dACC (Berntson and Khalsa, 2021; Craig, 2009, 2002). The insula, often referred to as the primary interoceptive cortex (Craig, 2003a, 2002), receives this Lamina I and NTS information along its posterior and dorsal regions, making the PI the primary site receiving direct interoceptive input (Barrett and Simmons, 2015; Craig, 2004; Dum et al., 2009; Kleckner et al., 2017; Uddin et al., 2017). In addition, bipolar recordings in the PI during an intracranial study of the HEP in humans have provided evidence the PI is a source of the HEP (Park et al., 2018). The current demonstration of site-specific (i.e., PI) modulation of the HEP within similar time windows identified in previous HEP studies provides additional evidence that the PI does indeed contribute to processing or generation of the observed HEP using surface EEG (Coll et al., 2021; Engelen et al., 2023; Park et al., 2018; Shao et al., 2011; Solcà et al., 2020).

The HEP looks to be malleable and is modulated by a range of attentional or arousal-based interoceptive tasks (Coll et al., 2021), including pain (Shao et al., 2011). For example, both experimentally-induced pain and chronic pain leads to a concomitant reduction in HEP amplitude that correlated with pain ratings or measures of interoceptive awareness (Shao et al., 2011; Solcà et al., 2020). One potential explanation of these results is the presence of pain, whether acute or chronic, overrides or attenuates interoceptive information due to the high saliency of pain. It is possible LIFU to the PI engaged a similar mechanism by directly inhibiting ascending interoceptive information. This would match prior literature that shows the exact same LIFU parameters are inhibitory across a range of brain targets (Legon et al., 2024, 2018b, 2018a, 2014; Strohman et al., 2024).

4.2. LIFU to the AI affected the HEP relative to dACC

We found no evidence for LIFU to the AI or dACC to affect the amplitude of the HEP compared to sham in the original time domain analysis. After controlling for the effects of pressure, the second time window (~655ms post R-peak) showed significant differences in the marginal means between dACC and AI conditions, but not between AI and PI. The dACC, similar to the PI, receives direct afferent cardiovascular information from the thalamus but through a distinct nucleus (MDvc) from what supplies the PI (VMpo) (Blomqvist and Evrard, 2024; Craig, 2003b, 2004; Dum et al., 2009). The AI, while not receiving direct afferent thalamic projections, communicates with the PI and dACC and serves as an integration hub which helps switch between large-scale brain networks and contributes to efferent visceromotor control (Barrett and Simmons, 2015; Frot et al., 2014; Khalsa et al., 2018; Kleckner et al., 2017; Seeley, 2019). Prior intracranial work has demonstrated the HEP can be recorded from the dACC and anterior insula (Park et al., 2018). One potential explanation for our results is that subregions of the AI and dACC have distinct structural and functional connectivity profiles, thus likely sub-serving different functions (Beckmann et al., 2009; Ghaziri et al., 2017; Han et al., 2019; Kleckner et al., 2017; Uddin, 2015). For example, studies attempting to disentangle AI vs dACC function suggest the AI signals the presence of significant events while the dACC updates internal attention in responses to the events (Han et al., 2019). It is therefore plausible that LIFU to the insula attenuated the saliency of the interoceptive signal to the dACC, represented as lower HEP amplitudes in the second time window, whereas the signal was unimpeded with LIFU to the dACC.

Interpreting these results requires additional work to comprehensively map the role of insular and cingulate subregions on the HEP. Other insular targets such as the mid-insula, often grouped with the AI, appear to play a distinct role in interoceptive processing (Adamic et al., 2024; Paulus and Khalsa, 2021). For the cingulate, our target was posterior to dACC sites identified both through intracranial and fMRI studies of the HEP, while other studies suggest even more posterior regions like the posterior cingulate may also encode the HEP (Park et al., 2018; Park and Blanke, 2019). Finally, a range of different tasks have produced changes in the HEP at various time points (Coll et al., 2021). It is unclear if there is some relationship between the task being performed and the time window of HEP modulation, and furthermore if modulation of the insula or cingulate during different tasks would lead to different effects on the HEP.

4.3. LIFU pressure partially accounts for the response to LIFU in humans

After accounting for pressure, results from the first time window demonstrated higher pressures in the brain were associated with greater changes in HEP amplitudes. Pressure was not associated with HEP changes in the second time window. Prior in-vitro and animal work has consistently shown dose-response relationships with LIFU where more energy produces stronger effects on the outcome of interest (Mohammadjavadi et al., 2019; Murphy et al., 2022; Tufail et al., 2010; Yoo et al., 2022). While previous human studies have reported estimated intracranial energy (Lee et al., 2016, 2015; Legon et al., 2023b; Strohman et al., 2024), the present investigation along with its companion behavioral study (In et al., 2024) provide early evidence of a dose-response to LIFU in humans on measures ranging from pain perception to neurophysiologic responses. Interestingly, in the present study the observed effects of pressure on HEP amplitudes in the first time window were similar for both PI and dACC conditions, but not for AI. This finding matches our hypotheses that the PI and dACC would have similar effects on HEP amplitudes due to both receiving direct afferent interoceptive information as opposed to the AI (Craig, 2004, 2002; Dum et al., 2009; Palma and Benarroch, 2014).

Dose-responses are a valuable measure for assessing causality in any human brain mapping study, and establishing a dose-response relationship for a desired effect at a specific neural target is critical for clinical applications of any non-invasive brain stimulation technique (Siddiqi et al., 2022). The differential effect of pressure in the present study helps build this evidence base for LIFU, although more work must be done to establish a minimal effective dose and optimal dosing.

4.4. Limitations and future directions

There are several limitations to this study. Although this proof-of-concept study demonstrated modulation of the HEP from LIFU to PI and AI, the reliance on a single electrode represents a limitation. Future work is needed to replicate the present findings across a larger number of scalp leads and in a larger sample size. Such studies could utilize data-driven cluster-based permutation approaches (Maris and Oostenveld, 2007) to identify electrode clusters showing maximal responses, which could in term support the application of source localization algorithms to analyze the underlying HEP generators. Our dose-response relationship was based on differences in estimated pressure as opposed to a pre-defined dosing strategy. Establishing clear dose-response relationships require studies explicitly exploring dosing, thus our findings of the effect of LIFU pressure should be considered preliminary. We did not have a pre-LIFU baseline data acquisition period, which would help rule out additional confounders not captured in our control analyses. Finally, our choice of targets within the AI, PI, and dACC may not necessarily reflect the optimal targets for investigating the role of these regions in the HEP. Therefore, the findings should not be taken as conclusive evidence of each region’s role in the HEP. More comprehensive mapping of insular and cingulate subregions and the use of more advanced brain stimulation targeting approaches such as personalized targets with resting state functional connectivity were not explored (Dijkstra et al., 2024), and this could be used in future studies to better answer questions about these regions’ contribution to the HEP.

5. CONCLUSIONS AND FUTURE DIRECTIONS

As a putative biomarker of brain-heart interaction, the HEP may be partially responsive to modulation of the insula by LIFU. Future work should investigate LIFU effects on the HEP using additional electrode leads, at a range of pre-defined doses, and in patient populations to evaluate clinical relevancy.

Supplementary Material

1
2
3

HIGHLIGHTS

  • Low-intensity focused ultrasound (LIFU) can target insular and cingulate cortex subregions.

  • Modulation of the anterior and posterior insula differentially decreased the heartbeat evoked potential (HEP).

  • The effects on the HEP were partially explained by LIFU pressure.

ACKNOWLEDGEMENTS

We would like to thank Jessica Florig for her assistance with data collection and data management.

FUNDING

This work was funded in part my grants awarded to WL from the National Institutes of Health (NIH R21AT012247–01).

Footnotes

DECLARATION OF INTEREST

The authors declare no conflicts of interest.

Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

REFERENCES

  1. Adamic EM, Teed AR, Avery JA, Cruz F de la, Khalsa SS. Hemispheric divergence of interoceptive processing across psychiatric disorders. eLife 2024;13. 10.7554/eLife.92820.1. [DOI] [PMC free article] [PubMed] [Google Scholar]
  2. Aubry J-F, Tanter M, Pernot M, Thomas J-L, Fink M. Experimental demonstration of noninvasive transskull adaptive focusing based on prior computed tomography scans. The Journal of the Acoustical Society of America 2003;113:84–93. 10.1121/1.1529663. [DOI] [PubMed] [Google Scholar]
  3. Aziz Q, Furlong PL, Barlow J, Hobson A, Alani S, Bancewicz J, et al. Topographic mapping of cortical potentials evoked by distension of the human proximal and distal oesophagus. Electroencephalogr Clin Neurophysiol 1995;96:219–28. 10.1016/0168-5597(94)00297-r. [DOI] [PubMed] [Google Scholar]
  4. Barrett LF, Simmons WK. Interoceptive predictions in the brain | Nature Reviews Neuroscience. Nature Reviews Neuroscience 2015;16:419–29. 10.1038/nrn3950. [DOI] [PMC free article] [PubMed] [Google Scholar]
  5. Beckmann M, Johansen-Berg H, Rushworth MFS. Connectivity-Based Parcellation of Human Cingulate Cortex and Its Relation to Functional Specialization. J Neurosci 2009;29:1175–90. 10.1523/JNEUROSCI.3328-08.2009. [DOI] [PMC free article] [PubMed] [Google Scholar]
  6. Beissner F, Meissner K, Bär K-J, Napadow V. The Autonomic Brain: An Activation Likelihood Estimation Meta-Analysis for Central Processing of Autonomic Function. J Neurosci 2013;33:10503–11. 10.1523/JNEUROSCI.1103-13.2013. [DOI] [PMC free article] [PubMed] [Google Scholar]
  7. Berntson GG, Khalsa SS. Neural Circuits of Interoception. Trends Neurosci 2021;44:17–28. 10.1016/j.tins.2020.09.011. [DOI] [PMC free article] [PubMed] [Google Scholar]
  8. Blomqvist A, Evrard HC. The thalamic projection of pain sensations to the posterior dorsal fundus in the insula: comment on Mandonnet et al. Pain 2024;165:e15–6. 10.1097/j.pain.0000000000003164. [DOI] [PMC free article] [PubMed] [Google Scholar]
  9. Bonaz B, Lane RD, Oshinsky ML, Kenny PJ, Sinha R, Mayer EA, et al. Diseases, Disorders, and Comorbidities of Interoception. Trends in Neurosciences 2021;44:39–51. 10.1016/j.tins.2020.09.009. [DOI] [PubMed] [Google Scholar]
  10. Coll M-P, Hobson H, Bird G, Murphy J. Systematic review and meta-analysis of the relationship between the heartbeat-evoked potential and interoception. Neuroscience & Biobehavioral Reviews 2021;122:190–200. 10.1016/j.neubiorev.2020.12.012. [DOI] [PubMed] [Google Scholar]
  11. Craig AD. A new view of pain as a homeostatic emotion. Trends in Neurosciences 2003a;26:303–7. 10.1016/S0166-2236(03)00123-1. [DOI] [PubMed] [Google Scholar]
  12. Craig AD. Pain mechanisms: Labeled lines versus convergence in central processing. Annual Review of Neuroscience 2003b;26:1–30. [DOI] [PubMed] [Google Scholar]
  13. Craig AD (Bud). How do you feel — now? The anterior insula and human awareness. Nat Rev Neurosci 2009;10:59–70. 10.1038/nrn2555. [DOI] [PubMed] [Google Scholar]
  14. Craig AD (Bud). Distribution of trigeminothalamic and spinothalamic lamina I terminations in the macaque monkey. Journal of Comparative Neurology 2004;477:119–48. 10.1002/cne.20240. [DOI] [PubMed] [Google Scholar]
  15. Craig AD (Bud). How do you feel? Interoception: the sense of the physiological condition of the body. Nat Rev Neurosci 2002;3:655–66. 10.1038/nrn894. [DOI] [PubMed] [Google Scholar]
  16. Craig ADB. Topographically organized projection to posterior insular cortex from the posterior portion of the ventral medial nucleus in the long-tailed macaque monkey. J Comp Neurol 2014;522:36–63. 10.1002/cne.23425. [DOI] [PMC free article] [PubMed] [Google Scholar]
  17. Darmani G, Bergmann TO, Butts Pauly K, Caskey CF, de Lecea L, Fomenko A, et al. Non-invasive transcranial ultrasound stimulation for neuromodulation. Clin Neurophysiol 2022;135:51–73. 10.1016/j.clinph.2021.12.010. [DOI] [PubMed] [Google Scholar]
  18. Di Lernia D, Serino S, Riva G. Pain in the body. Altered interoception in chronic pain conditions: A systematic review. Neuroscience & Biobehavioral Reviews 2016;71:328–41. 10.1016/j.neubiorev.2016.09.015. [DOI] [PubMed] [Google Scholar]
  19. Dijkstra ESA, Frandsen SB, van Dijk H, Duecker F, Taylor JJ, Sack AT, et al. Probing prefrontalsgACC connectivity using TMS-induced heart–brain coupling. Nat Mental Health 2024:1–9. 10.1038/s44220-024-00248-8. [DOI] [Google Scholar]
  20. Dum RP, Levinthal DJ, Strick PL. The Spinothalamic System Targets Motor and Sensory Areas in the Cerebral Cortex of Monkeys. J Neurosci 2009;29:14223–35. 10.1523/JNEUROSCI.3398-09.2009. [DOI] [PMC free article] [PubMed] [Google Scholar]
  21. Engelen T, Solcà M, Tallon-Baudry C. Interoceptive rhythms in the brain. Nat Neurosci 2023:1–15. 10.1038/s41593-023-01425-1. [DOI] [PubMed] [Google Scholar]
  22. Frot M, Faillenot I, Mauguière F. Processing of nociceptive input from posterior to anterior insula in humans. Human Brain Mapping 2014;35:5486–99. 10.1002/hbm.22565. [DOI] [PMC free article] [PubMed] [Google Scholar]
  23. Gasparini S, Howland JM, Thatcher AJ, Geerling JC. Central afferents to the nucleus of the solitary tract in rats and mice. Journal of Comparative Neurology 2020;528:2708–28. 10.1002/cne.24927. [DOI] [PMC free article] [PubMed] [Google Scholar]
  24. Ghaziri J, Tucholka A, Girard G, Houde J-C, Boucher O, Gilbert G, et al. The Corticocortical Structural Connectivity of the Human Insula. Cerebral Cortex 2017;27:1216–28. 10.1093/cercor/bhv308. [DOI] [PubMed] [Google Scholar]
  25. Gray MA, Minati L, Paoletti G, Critchley HD. Baroreceptor activation attenuates attentional effects on pain-evoked potentials. Pain 2010;151:853–61. 10.1016/j.pain.2010.09.028. [DOI] [PMC free article] [PubMed] [Google Scholar]
  26. Han SW, Eaton HP, Marois R. Functional Fractionation of the Cingulo-opercular Network: Alerting Insula and Updating Cingulate. Cereb Cortex 2019;29:2624–38. 10.1093/cercor/bhy130. [DOI] [PMC free article] [PubMed] [Google Scholar]
  27. Hobday DI, Hobson AR, Sarkar S, Furlong PL, Thompson DG, Aziz Q. Cortical processing of human gut sensation: an evoked potential study. Am J Physiol Gastrointest Liver Physiol 2002;283:G335–339. 10.1152/ajpgi.00230.2001. [DOI] [PubMed] [Google Scholar]
  28. Hobson AR, Khan RW, Sarkar S, Furlong PL, Aziz Q. Development of esophageal hypersensitivity following experimental duodenal acidification. Am J Gastroenterol 2004;99:813–20. 10.1111/j.1572-0241.2004.04167.x. [DOI] [PubMed] [Google Scholar]
  29. In A, Strohman A, Payne B, Legon W. Low-intensity focused ultrasound to the insula and dorsal anterior cingulate has site-specific and pressure dependent effects on pain during measures of central sensitization 2024:2024.01.10.575098. 10.1101/2024.01.10.575098. [DOI] [Google Scholar]
  30. Kemp AH, Brunoni AR, Santos IS, Nunes MA, Dantas EM, Carvalho de Figueiredo R, et al. Effects of Depression, Anxiety, Comorbidity, and Antidepressants on Resting-State Heart Rate and Its Variability: An ELSA-Brasil Cohort Baseline Study. AJP 2014;171:1328–34. 10.1176/appi.ajp.2014.13121605. [DOI] [PubMed] [Google Scholar]
  31. Kemp AH, Quintana DS, Gray MA, Felmingham KL, Brown K, Gatt JM. Impact of Depression and Antidepressant Treatment on Heart Rate Variability: A Review and Meta-Analysis.Biological Psychiatry 2010;67:1067–74. 10.1016/j.biopsych.2009.12.012. [DOI] [PubMed] [Google Scholar]
  32. Kern M, Aertsen A, Schulze-Bonhage A, Ball T. Heart cycle-related effects on event-related potentials, spectral power changes, and connectivity patterns in the human ECoG. Neuroimage 2013;81:178–90. 10.1016/j.neuroimage.2013.05.042. [DOI] [PubMed] [Google Scholar]
  33. Khalsa SS, Adolphs R, Cameron OG, Critchley HD, Davenport PW, Feinstein JS, et al. Interoception and Mental Health: A Roadmap. Biological Psychiatry: Cognitive Neuroscience and Neuroimaging 2018;3:501–13. 10.1016/j.bpsc.2017.12.004. [DOI] [PMC free article] [PubMed] [Google Scholar]
  34. Kleckner IR, Zhang J, Touroutoglou A, Chanes L, Xia C, Simmons WK, et al. Evidence for a large-scale brain system supporting allostasis and interoception in humans. Nat Hum Behav 2017;1:1–14. 10.1038/s41562-017-0069. [DOI] [PMC free article] [PubMed] [Google Scholar]
  35. Lee W, Kim H, Jung Y, Song I-U, Chung YA, Yoo S-S. Image-Guided Transcranial Focused Ultrasound Stimulates Human Primary Somatosensory Cortex. Sci Rep 2015;5:8743. 10.1038/srep08743. [DOI] [PMC free article] [PubMed] [Google Scholar]
  36. Lee W, Kim H-C, Jung Y, Chung YA, Song I-U, Lee J-H, et al. Transcranial focused ultrasound stimulation of human primary visual cortex. Sci Rep 2016;6:34026. 10.1038/srep34026. [DOI] [PMC free article] [PubMed] [Google Scholar]
  37. Legon W, Adams S, Bansal P, Patel PD, Hobbs L, Ai L, et al. A retrospective qualitative report of symptoms and safety from transcranial focused ultrasound for neuromodulation in humans. Scientific Reports 2020;10:1–10. [DOI] [PMC free article] [PubMed] [Google Scholar]
  38. Legon W, Ai L, Bansal P, Mueller JK. Neuromodulation with single‐element transcranial focused ultrasound in human thalamus. Human Brain Mapping 2018a;39:1995–2006. [DOI] [PMC free article] [PubMed] [Google Scholar]
  39. Legon W, Bansal P, Tyshynsky R, Ai L, Mueller JK. Transcranial focused ultrasound neuromodulation of the human primary motor cortex. Scientific Reports 2018b;8:1–14. [DOI] [PMC free article] [PubMed] [Google Scholar]
  40. Legon W, Sato TF, Opitz A, Mueller J, Barbour A, Williams A, et al. Transcranial focused ultrasound modulates the activity of primary somatosensory cortex in humans. Nat Neurosci 2014;17:322–9. 10.1038/nn.3620. [DOI] [PubMed] [Google Scholar]
  41. Legon W, Strohman A, In A, Payne B. Noninvasive neuromodulation of subregions of the human insula differentially affect pain processing and heart-rate variability: a within-subjects pseudo-randomized trial. PAIN 2024. 10.1097/j.pain.0000000000003171. [DOI] [PMC free article] [PubMed] [Google Scholar]
  42. Legon W, Strohman A, In A, Stebbins K, Payne B. Non-invasive neuromodulation of subregions of the human insula differentially affect pain processing and heart-rate variability. bioRxiv 2023b:2023.05.05.539593. 10.1101/2023.05.05.539593. [DOI] [PMC free article] [PubMed] [Google Scholar]
  43. Liang W, Guo H, Mittelstein DR, Shapiro MG, Shimojo S, Shehata MH. Auditory Mondrian masks the airborne-auditory artifact of focused ultrasound stimulation in humans. Brain Stimulation: Basic, Translational, and Clinical Research in Neuromodulation 2023;0. 10.1016/j.brs.2023.03.002. [DOI] [PMC free article] [PubMed] [Google Scholar]
  44. Mandonnet V, Obaid S, Descoteaux M, St-Onge E, Devaux B, Levé C, et al. Electrostimulation of the white matter of the posterior insula and medial operculum: perception of vibrations, heat, and pain. Pain 2024;165:565–72. 10.1097/j.pain.0000000000003069. [DOI] [PubMed] [Google Scholar]
  45. Maris E, Oostenveld R. Nonparametric statistical testing of EEG- and MEG-data. J Neurosci Methods 2007;164:177–90. 10.1016/j.jneumeth.2007.03.024. [DOI] [PubMed] [Google Scholar]
  46. Marquet F, Boch A-L, Pernot M, Montaldo G, Seilhean D, Fink M, et al. Non-invasive ultrasonic surgery of the brain in non-human primates. The Journal of the Acoustical Society of America 2013;134:1632–9. 10.1121/1.4812888. [DOI] [PubMed] [Google Scholar]
  47. Mazzola L, Mauguière F, Chouchou F. Central control of cardiac activity as assessed by intra-cerebral recordings and stimulations. Neurophysiologie Clinique 2023;53:102849. 10.1016/j.neucli.2023.102849. [DOI] [PubMed] [Google Scholar]
  48. Mohammadjavadi M, Ye PP, Xia A, Brown J, Popelka G, Pauly KB. Elimination of peripheral auditory pathway activation does not affect motor responses from ultrasound neuromodulation. Brain Stimulation 2019;12:901–10. 10.1016/j.brs.2019.03.005. [DOI] [PMC free article] [PubMed] [Google Scholar]
  49. Mueller J, Legon W, Opitz A, Sato TF, Tyler WJ. Transcranial Focused Ultrasound Modulates Intrinsic and Evoked EEG Dynamics. Brain Stimulation 2014;7:900–8. 10.1016/j.brs.2014.08.008. [DOI] [PubMed] [Google Scholar]
  50. Mueller JK, Ai L, Bansal P, Legon W. Numerical evaluation of the skull for human neuromodulation with transcranial focused ultrasound. J Neural Eng 2017;14:066012. 10.1088/1741-2552/aa843e. [DOI] [PubMed] [Google Scholar]
  51. Mueller JK, Ai L, Bansal P, Legon W. Computational exploration of wave propagation and heating from transcranial focused ultrasound for neuromodulation. Journal of Neural Engineering 2016;13:056002. [DOI] [PubMed] [Google Scholar]
  52. Murphy KR, Farrell JS, Gomez JL, Stedman QG, Li N, Leung SA, et al. A tool for monitoring cell type–specific focused ultrasound neuromodulation and control of chronic epilepsy. Proceedings of the National Academy of Sciences 2022;119:e2206828119. 10.1073/pnas.2206828119. [DOI] [PMC free article] [PubMed] [Google Scholar]
  53. Nakajima K, Osada T, Ogawa A, Tanaka M, Oka S, Kamagata K, et al. A causal role of anterior prefrontal-putamen circuit for response inhibition revealed by transcranial ultrasound stimulation in humans. Cell Reports 2022;40:111197. 10.1016/j.celrep.2022.111197. [DOI] [PubMed] [Google Scholar]
  54. Palma J-A, Benarroch EE. Neural control of the heart: Recent concepts and clinical correlations. Neurology 2014;83:261–71. 10.1212/WNL.0000000000000605. [DOI] [PubMed] [Google Scholar]
  55. Pang J, Tang X, Li H, Hu Q, Cui H, Zhang L, et al. Altered Interoceptive Processing in Generalized Anxiety Disorder—A Heartbeat-Evoked Potential Research. Frontiers in Psychiatry 2019;10. [DOI] [PMC free article] [PubMed] [Google Scholar]
  56. Park H-D, Bernasconi F, Salomon R, Tallon-Baudry C, Spinelli L, Seeck M, et al. Neural Sources and Underlying Mechanisms of Neural Responses to Heartbeats, and their Role in Bodily Self-consciousness: An Intracranial EEG Study. Cerebral Cortex 2018;28:2351–64. 10.1093/cercor/bhx136. [DOI] [PubMed] [Google Scholar]
  57. Park H-D, Blanke O Heartbeat-evoked cortical responses: Underlying mechanisms, functional roles, and methodological considerations. NeuroImage 2019;197:502–11. 10.1016/j.neuroimage.2019.04.081. [DOI] [PubMed] [Google Scholar]
  58. Paulus MP, Khalsa SS. When You Don’t Feel Right Inside: Homeostatic Dysregulation and the Mid-Insular Cortex in Psychiatric Disorders. AJP 2021;178:683–5. 10.1176/appi.ajp.2021.21060622. [DOI] [PubMed] [Google Scholar]
  59. Petzschner FH, Weber LA, Wellstein KV, Paolini G, Do CT, Stephan KE. Focus of attention modulates the heartbeat evoked potential. NeuroImage 2019;186:595–606. 10.1016/j.neuroimage.2018.11.037. [DOI] [PubMed] [Google Scholar]
  60. Quadt L, Critchley HD, Garfinkel SN. The neurobiology of interoception in health and disease. Annals of the New York Academy of Sciences 2018;1428:112–28. 10.1111/nyas.13915. [DOI] [PubMed] [Google Scholar]
  61. Sarica C, Nankoo J-F, Fomenko A, Grippe TC, Yamamoto K, Samuel N, et al. Human Studies of Transcranial Ultrasound neuromodulation: A systematic review of effectiveness and safety. Brain Stimulation 2022;15:737–46. 10.1016/j.brs.2022.05.002. [DOI] [PubMed] [Google Scholar]
  62. Schandry R, Sparrer B, Weitkunat R. From the heart to the brain: A study of heartbeat contingent scalp potentials. International Journal of Neuroscience 1986;30:261–75. 10.3109/00207458608985677. [DOI] [PubMed] [Google Scholar]
  63. Seeley WW. The Salience Network: A Neural System for Perceiving and Responding to Homeostatic Demands. J Neurosci 2019;39:9878–82. 10.1523/JNEUROSCI.1138-17.2019. [DOI] [PMC free article] [PubMed] [Google Scholar]
  64. Seth AK. Interoceptive inference, emotion, and the embodied self. Trends in Cognitive Sciences 2013;17:565–73. 10.1016/j.tics.2013.09.007. [DOI] [PubMed] [Google Scholar]
  65. Shaffer F, Ginsberg JP. An Overview of Heart Rate Variability Metrics and Norms. Frontiers in Public Health 2017;5. [DOI] [PMC free article] [PubMed] [Google Scholar]
  66. Shao S, Shen K, Wilder-Smith EPV, Li X. Effect of pain perception on the heartbeat evoked potential. Clinical Neurophysiology 2011;122:1838–45. 10.1016/j.clinph.2011.02.014. [DOI] [PubMed] [Google Scholar]
  67. Siddiqi SH, Kording KP, Parvizi J, Fox MD. Causal mapping of human brain function. Nat Rev Neurosci 2022;23:361–75. 10.1038/s41583-022-00583-8. [DOI] [PMC free article] [PubMed] [Google Scholar]
  68. Siddiqi SH, Schaper FLWVJ, Horn A, Hsu J, Padmanabhan JL, Brodtmann A, et al. Brain stimulation and brain lesions converge on common causal circuits in neuropsychiatric disease. Nat Hum Behav 2021;5:1707–16. 10.1038/s41562-021-01161-1. [DOI] [PMC free article] [PubMed] [Google Scholar]
  69. Solcà M, Park H-D, Bernasconi F, Blanke O. Behavioral and neurophysiological evidence for altered interoceptive bodily processing in chronic pain. NeuroImage 2020;217:116902. 10.1016/j.neuroimage.2020.116902. [DOI] [PubMed] [Google Scholar]
  70. Strohman A, In A, Stebbins K, Legon W. Evaluation of a Novel Acoustic Coupling Medium for Human Low-Intensity Focused Ultrasound Neuromodulation Applications. Ultrasound in Medicine & Biology 2023. 10.1016/j.ultrasmedbio.2023.02.003. [DOI] [PubMed] [Google Scholar]
  71. Strohman A, Payne B, In A, Stebbins K, Legon W. Low-intensity focused ultrasound to the human dorsal anterior cingulate attenuates acute pain perception and autonomic responses. J Neurosci 2024. 10.1523/JNEUROSCI.1011-23.2023. [DOI] [PMC free article] [PubMed] [Google Scholar]
  72. Tanner D, Morgan-Short K, Luck SJ. How inappropriate high-pass filters can produce artifactual effects and incorrect conclusions in ERP studies of language and cognition. Psychophysiology 2015;52:997–1009. 10.1111/psyp.12437. [DOI] [PMC free article] [PubMed] [Google Scholar]
  73. Taylor JJ, Lin C, Talmasov D, Ferguson MA, Schaper FLWVJ, Jiang J, et al. A transdiagnostic network for psychiatric illness derived from atrophy and lesions. Nat Hum Behav 2023:1–10. 10.1038/s41562-022-01501-9. [DOI] [PMC free article] [PubMed] [Google Scholar]
  74. Terhaar J, Viola FC, Bär K-J, Debener S. Heartbeat evoked potentials mirror altered body perception in depressed patients. Clinical Neurophysiology 2012;123:1950–7. 10.1016/j.clinph.2012.02.086. [DOI] [PubMed] [Google Scholar]
  75. Treeby BE, Cox BT. k-Wave: MATLAB toolbox for the simulation and reconstruction of photoacoustic wave fields. JBO 2010;15:021314. 10.1117/1.3360308. [DOI] [PubMed] [Google Scholar]
  76. Tufail Y, Matyushov A, Baldwin N, Tauchmann ML, Georges J, Yoshihiro A, et al. Transcranial Pulsed Ultrasound Stimulates Intact Brain Circuits. Neuron 2010;66:681–94. 10.1016/j.neuron.2010.05.008. [DOI] [PubMed] [Google Scholar]
  77. Tyler WJ, Lani SW, Hwang GM. Ultrasonic modulation of neural circuit activity. Current Opinion in Neurobiology 2018;50:222–31. 10.1016/j.conb.2018.04.011. [DOI] [PubMed] [Google Scholar]
  78. Uddin LQ. Salience processing and insular cortical function and dysfunction | Nature Reviews Neuroscience. Nature Reviews Neuroscience 2015;16:55–61. 10.1038/nrn3857. [DOI] [PubMed] [Google Scholar]
  79. Uddin LQ, Nomi JS, Hébert-Seropian B, Ghaziri J, Boucher O. Structure and Function of the Human Insula: Journal of Clinical Neurophysiology 2017;34:300–6. 10.1097/WNP.0000000000000377. [DOI] [PMC free article] [PubMed] [Google Scholar]
  80. Verdonk C, Teed AR, White EJ, Ren X, Stewart JL, Paulus MP, et al. Heartbeat-evoked neural response abnormalities in generalized anxiety disorder during peripheral adrenergic stimulation. Neuropsychopharmacol 2024:1–9. 10.1038/s41386-024-01806-5. [DOI] [PMC free article] [PubMed] [Google Scholar]
  81. Wager TD, Atlas LY, Lindquist MA, Roy M, Woo C-W, Kross E. An fMRI-Based Neurologic Signature of Physical Pain. N Engl J Med 2013a;368:1388–97. 10.1056/NEJMoa1204471. [DOI] [PMC free article] [PubMed] [Google Scholar]
  82. Wager TD, Atlas LY, Lindquist MA, Roy M, Woo C-W, Kross E. An fMRI-Based Neurologic Signature of Physical Pain. N Engl J Med 2013b;368:1388–97. 10.1056/NEJMoa1204471. [DOI] [PMC free article] [PubMed] [Google Scholar]
  83. Widmann A, Schröger E, Maess B. Digital filter design for electrophysiological data – a practical approach. Journal of Neuroscience Methods 2015;250:34–46. 10.1016/j.jneumeth.2014.08.002. [DOI] [PubMed] [Google Scholar]
  84. Yaakub SN, White TA, Roberts J, Martin E, Verhagen L, Stagg CJ, et al. Transcranial focused ultrasound-mediated neurochemical and functional connectivity changes in deep cortical regions in humans | Nature Communications. Nature Communications 2023;14:5318. 10.1038/s41467-023-40998-0. [DOI] [PMC free article] [PubMed] [Google Scholar]
  85. Yarkoni T, Poldrack RA, Nichols TE, Van Essen DC, Wager TD. Large-scale automated synthesis of human functional neuroimaging data. Nat Methods 2011;8:665–70. 10.1038/nmeth.1635. [DOI] [PMC free article] [PubMed] [Google Scholar]
  86. Yoo S, Mittelstein DR, Hurt RC, Lacroix J, Shapiro MG. Focused ultrasound excites cortical neurons via mechanosensitive calcium accumulation and ion channel amplification. Nat Commun 2022;13:493. 10.1038/s41467-022-28040-1. [DOI] [PMC free article] [PubMed] [Google Scholar]
  87. Zebhauser PT, Hohn VD, Ploner M. Resting-state electroencephalography and magnetoencephalography as biomarkers of chronic pain: a systematic review. Pain 2023;164:1200–21. 10.1097/j.pain.0000000000002825. [DOI] [PMC free article] [PubMed] [Google Scholar]
  88. Zeng K, Darmani G, Fomenko A, Xia X, Tran S, Nankoo J-F, et al. Induction of Human Motor Cortex Plasticity by Theta Burst Transcranial Ultrasound Stimulation. Annals of Neurology 2022;91:238–52. 10.1002/ana.26294. [DOI] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

1
NIHMS2024333-supplement-1.docx (12.2KB, docx)
2
NIHMS2024333-supplement-2.pdf (428.2KB, pdf)
3
NIHMS2024333-supplement-3.pdf (107.7KB, pdf)

RESOURCES