Transcranial Focused Ultrasound Stimulation Targeting White Matter Inhibits Seizures in a Rat Model of Epilepsy

Abstract

Transcranial Focused Ultrasound Stimulation (tFUS) is a promising non-invasive technique capable of modulating brain activity with high spatial precision. However, its efficacy for seizure suppression requires further exploration. This study aims to address whether tFUS of white matter can suppress seizures non-invasively. Repeated injections of a 4-Aminopyridine (4-AP) cocktail into the right somatosensory cortex (S1) induced cortical seizures in male rats under anesthesia with recording of both EEG and intracranial signals. Approximately one hour of tFUS was applied to the corpus callosum (CC) using a 128-element random-array transducer with 20 ms pulse duration, 1 Hz pulse repetition frequency, 2% duty cycle, and ~127 kPa pressure. Another 2–3 hours were used to assess post-stimulation effects. Seizure duration, seizure count, percent time in seizure, and inter-seizure interval were compared to a sham control for quantifying efficacy. The absolute frequency power, asymmetry index (AI), and phase lag index (PLI) were calculated to analyze brain activity changes induced by tFUS. CC tFUS can significantly reduce percent time in seizure, seizure duration, and seizure count, as well as increase inter-seizure interval. These effects extended up to 2 hours post-stimulation. We also observed a decrease in absolute power of the beta band and changes in the brain network, as evidenced by a decrease in synchronization and an improvement in interhemispheric balance. Our study is the first to show that white matter tFUS can significantly suppress seizures with a lasting post stimulation effect, potentially providing a safer alternative for drug-resistant epilepsy patients.

I. Introduction

EPILEPSY is a pervasive neurological disorder associated with distressing symptoms, serious comorbidities, and reduced quality of life [1]. Of the estimated 65 million people affected worldwide [2], approximately one-third have drug-resistant epilepsy (DRE) and are ineligible for standard pharmacological treatments [3], [4], [5], [6]. DRE patients may be eligible for resection [7]; however, even if the onset zone(s) are well-defined and located in a non-crucial brain area, only about 50% of patients who opt for resection achieve seizure freedom [8], [9].

Non-invasive neuromodulation is an alternative to pharmacological and surgical treatments. Both transcranial magnetic stimulation (TMS) and transcranial direct current stimulation (tDCS) have been studied as potential non-invasive therapies, however, they have limited spatial resolution [3], [4], [10]. Transcranial focused ultrasound stimulation (tFUS), which uses sound waves to non-invasively modulate neuronal activity [11], [12], [13], [14], offers high spatial specificity and deep brain penetration. The safety and efficacy of low-intensity tFUS for epilepsy treatment have been suggested by preclinical and clinical studies. In rodent models of epilepsy, tFUS has been shown to decrease seizure count [15] and duration [15], [16] as well as suppress epileptic spikes and bursts [3], [10], [12], [17], [18], [19], [20]. Furthermore, tFUS can delay status epilepticus [20] and improve survival rates [16]. Several behavioral studies conducted on epileptic rodents have shown that tFUS can improve sociability [20], [21], limb usage [21], and depression [20]. However, there are conflicting results on whether tFUS affects anxiety in epileptic rodents, with some studies showing improvement [20], [22] and others showing no effect [19]. In terms of safety, low-intensity tFUS applied to epileptic rodents does not induce tissue damage or inflammation [6], [16], [18], [21], and can provide neuroprotective effects [4], [21], [22], [23]. In the limited number of clinical trials applying tFUS to DRE patients, researchers observed no histological tissue damage and minor adverse events (e.g., scalp heating) [24], [25], [26], [27]. Of these studies, two groups have assessed the efficacy of tFUS for seizure suppression. Lee et al. found that tFUS reduced seizures in two patients and increased seizures in a third; the remaining three patients had no observable seizures during the study [26]. Bubrick et al. observed a reduction in seizure frequency for all six patients, with 5 of them experiencing significant improvement [27]. More clinical trials applying tFUS to DRE patients are currently underway [8].

Despite these promising results, the efficacy of tFUS for seizure suppression remains suboptimal, and continued research is required to enhance the therapeutic effect. Identifying a stimulation target that maximizes the inhibitory effect of tFUS could improve efficacy. All of the aforementioned studies applying tFUS to rodent models of epilepsy target the thalamus, hippocampus, or cortex. We propose to test a novel target for tFUS, namely the corpus callosum (CC) (the largest white matter tract in the brain) [28], based on its success in invasive neuromodulation [7], [9], [29], [30], [31], [32]. White matter stimulation is advantageous because of its potential to activate large brain regions simultaneously, particularly in cases where seizure foci are difficult to localize or the seizure spreads too quickly for focal treatment [29], [32]. Invasive low-frequency (1–20 Hz) stimulation (LFS) of the CC has proven effective at reducing seizures in rodent models of epilepsy [7], [9]. Furthermore, invasive stimulation of the fornix in a human clinical trial reduced seizure odds by 92% two days after stimulation delivery, demonstrating the clinical feasibility of this approach [29]. Therefore, white matter may be a promising new target for tFUS.

The purpose of this study is to test the efficacy of white matter tFUS as a potential non-invasive treatment for epilepsy. We induced seizures in male rats using 4-aminopyridine (4-AP) and applied stimulation to the CC segment with fibers innervating the seizure onset zone. We analyze the seizure characteristics and brain signals to quantify the suppressive effect of tFUS stimulation and compare these results to a sham group without tFUS.

II. Materials and Methods

A. Animals and Surgical Procedure.

This study was conducted according to a protocol approved by the Institutional Animal Care and Use Committee (IACUC) of Carnegie Mellon University—approval: IPROTO202300000003, following the Public Health Service Policy on Humane Care and Use of Laboratory Animals. 16 adult male Sprague-Dawley rats (Envigo; 250–350 g) aging 3–6 months were used in the epilepsy study, 10 in the experimental group and 6 in the sham group. An additional 7 rats (23 total) were included in a mechanism study. The animals were anesthetized using isoflurane (4% induction, 1–3% maintenance). Heart rate, respiratory rate, and body temperature were monitored throughout the surgery and subsequent experiment to ensure proper anesthetic depth. A stereotaxic frame was used to secure the head, and after confirming no toe pinch reflex, an incision was made along the rostral-caudal axis, and the scalp was removed to gain access to the skull. Two burr holes were created over the somatosensory cortex (S1) on each hemisphere. To account for variations from the rat brain atlas, the exact burr hole positioning was scaled according to the empirical distance between bregma and lambda.

A 32-channel EEG probe (NeuroNexus HC32 Rat Functional) was positioned on the surface of the skull to record brain-wide activity, which is similar to previous work [33]. A microsyringe was then inserted into the right S1 (AP: −2 mm, ML: 2.5 mm, DV: −1.5 mm) at an angle of 30 degrees relative to the vertical axis, and a 64-channel intracranial electrode array (NeuroNexus, A1×64-Edge-6mm-20–177) was inserted into the left S1 (AP: −2 mm, ML: −2.5 mm, DV: −1.5 mm) at 37 degrees relative to the vertical axis. These injections were done at an angle to allow for the placement of the tFUS transducer over the targeted region of the CC. Fig 1A provides a visual of the experimental setup.

Fig. 1.

Experimental setup and design. A, Depiction of transducer’s placement relative to stereotaxic frame. Excerpt contains the top view of the skull before placement of the transducer. The EEG probe is placed directly on the skull, aligned with bregma. The EEG electrode encircled by the blue/orange ring represents the selected channel data used to label seizures from the ipsilateral/contralateral hemisphere relative to the drug injection. The intracranial electrode is inserted into the left somatosensory cortex (S1), and the micro-syringe is inserted into the right S1. B, Transcranial focused ultrasound stimulation (tFUS) waveform and parameters used during the experimental condition. The pulse duration (PD), pulse repetition frequency (PRF = 1/PRP, pulse repetition period), duty cycle (DC), and in situ pressure were set to 20 ms, 1 Hz, 2%, and ~127 kPa, respectively. C-D, Pressure distribution of the tFUS transducer in free water with a rat skull, as seen from the frontal (C) and transverse (D) planes. E, Timeline of the experimental design. After performing target confirmation, the seizure model was generated with repeated injections of the 4-AP cocktail. Stimulation was either applied (tFUS group) or not (sham) during the condition time block, and then another three post-stimulation recordings were collected.

B. Epilepsy Model.

Focal cortical seizures were induced through repeated injections of a 4-AP cocktail solution consisting of 30 mmol/L 4-AP, 1.2 mmol/L CaCl₂, and 0.6 mmol/L MgSO₄ [7], [9]. 1.2 μL injections were administered into the right S1 at 2 μL/min and repeated at approximately 62.5-minute intervals to match the experimental time blocks illustrated in Fig. 1E. Previous studies showed fibers crossing between hemispheres can promote synchronization and seizure generation [7], which allowed us to use the LFP activity in the contralateral hemisphere as a proxy for the activity observed at the seizure onset zone (ipsilateral hemisphere) during target confirmation.

C. Seizure Identification and Quantification.

All data were recorded using a commercial multi-channel neural signal acquisition system (Tucker-Davis Technologies, Alachua, FL, USA) at a sampling rate of 3051.8 Hz; a 60 Hz notch filter was applied to remove powerline noise and signals were bandpass filtered between 3–300 Hz. Independent component analysis (ICA) and common average reference (CAR) were performed on the EEG signal to remove artifacts (e.g., respiratory/heart) and common noise, respectively. Seizures were manually labeled as time periods where the EEG signal’s summed power between 30–80 Hz exceeded a manually set threshold for at least 10 seconds [34]. Labeled seizures occurring less than 10 seconds apart were combined. The hemispheres were labeled independently using the EEG electrode closest to that side’s respective burr hole. The experimenter labeling the seizures was blinded to the time block from which the data originated (e.g., baseline vs post-stimulation) to mitigate bias.

Using the labeled EEG data from each hemisphere, the following four metrics were calculated and averaged among rats to quantify efficacy: average Seizure Duration (length of the seizures), Seizure Counts (number of seizures), Percent Time in Seizure (time spent seizing divided by total time considered), and Inter-Seizure Interval (time between neighboring seizures). Seizures that either started before or continued after the recording were excluded from the duration calculations. For the Inter-Seizure Interval calculations, two rats were excluded because they had conditions without seizure occurrence.

D. Transcranial Focused Ultrasound Stimulation (tFUS).

tFUS was applied using a 128-element random array transducer (H276, f₀ = 1.5 MHz) and corresponding Verasonics Vantage 128 ultrasound system [35]. Before conducting the in-vivo experiments, the transducer’s pressure distribution was approximated ex-vivo. The resulting acoustic pressure field is depicted in Fig. 1C-D. Details of the scanning process are provided in Supplementary Materials.

Because CC stimulation is spatially selective [9], accurate targeting is crucial for achieving a therapeutic effect in-vivo. To this end, evoked potentials (EPs) recorded from the intracranial electrode in the left S1 were employed as a means of target confirmation for tFUS. Details on the target confirmation process are provided in Supplementary Materials and Fig. S2.

After finalizing the stimulation coordinates, the first drug injection was administered to the right S1. Seizures typically developed within the first 10–20 minutes, but seizure generation was monitored for ~62.5 minutes to ensure model stability. If there were less than 5 seizures in the first recording, we either increased the injection volume by 0.3 uL or lowered the isoflurane level, and another ~62.5 minutes was monitored before the baseline recording. During the condition time block, either tFUS was delivered to the CC or no stimulation was applied (sham). The PD, PRF, DC, and pressure were set to 20 ms, 1 Hz, 2%, and ~127 kPa, respectively (Fig. 1B). A total of 3,750 pulses of stimulation (~62.5 minutes) were applied to the CC in the experimental group. Another ~3 hours of recordings were collected to monitor potential post-stimulation effects (stimulation condition) or the model’s stability over time (sham). Fig. 1E contains a timeline of the experimental design.

E. tFUS on Brain Activity.

To analyze power changes induced by tFUS stimulation, the EEG power spectral density (PSD) was calculated within the theta (5–8 Hz), alpha (8–12 Hz), beta (12–30 Hz), and gamma (30–80 Hz) bands using the Fourier transform. The PSD values were determined using the EEG electrode closest to the burr hole within each hemisphere. We then calculated the absolute power of each frequency band by dividing the frequency-specific PSD values by the total PSD of entire spectrum. Furthermore, the asymmetry index (AI), which represents the imbalance between hemispheres, was calculated using Eq. 1 for each of the aforementioned frequency bands by taking the relative difference in PSD between the ipsilateral and contralateral S1 [36], [37]. To assess tFUS-induced changes in interhemispheric imbalance relative to the baseline seizure condition—rather than natural hemispheric asymmetry [38] or subject-dependent variability—the AI for each time block was normalized to the group’s baseline AI.

$$AI=\frac{{\mathit{ipsilateral}}_{PSD}-{\mathit{contralateral}}_{PSD}}{{\mathit{ipsilateral}}_{PSD}+{\mathit{contralateral}}_{PSD}}$$ (1)

To assess functional connectivity changes between hemispheres, we calculated the phase lag index (PLI) within the theta, alpha, beta, and gamma bands [6]. Given two EEG signals $x\left(t\right)$ and $y\left(t\right)$ (one from each hemisphere), the instantaneous phase difference $\mathrm{\Delta }\theta \left(t\right)={\theta }_x\left(t\right)-{\theta }_y\left(t\right)$ was computed at each time point using the Hilbert transform. The PLI was defined as

$$PLI=|<sign(sin(\Delta \theta (t\left)\right))>|$$ (2)

where $<•>$ denotes the average over time. The PLI was used to indicate whether the activity in one hemisphere led or lagged behind the other hemisphere. PLI ranges between 0 and 1, with low values suggesting weak functional connectivity and high values suggesting strong functional connectivity. We explored how tFUS affected wild-type rats by extracting 200 EPs from the target confirmation procedure and calculating the PLI of the following three-time segments: pre-tFUS (20 ms before stimulation onset), tFUS (20 ms of stimulation, and post-tFUS (20–200 ms after stimulation). The tFUS and post-tFUS values were normalized to the pre-tFUS PLI to better analyze changes induced by tFUS. A similar procedure was conducted on the baseline recording taken during target confirmation, which served as a resting state control (sham). We also investigated how CC tFUS affected brain connectivity in epileptic rats by comparing the seizures during baseline, condition, and post-stimulation time blocks. The PLI of each frequency band was normalized to the baseline seizure recording within the experimental group or sham group to highlight changes induced by tFUS. To explore whether CC stimulation specifically affected the S1 or on the entire brain, we separated the EEG channels into S1 regions and non-S1 regions (Fig. 5A) based on the placement of the EEG probes and rat atlas [39]. The PLIs between left S1 region and right S1 region, as well as between the left non-S1 region and right non-S1 region, were computed and compared.

Fig. 5.

Evoked potentials (EPs) induced by tFUS and corresponding brain connectivity changes in wild-type rats. A, Schematic of EEG probe, with black ellipses distinguishing the EEG electrodes covering the S1 areas. B, EPs extracted from LFPs are normalized to a 20 ms pre-tFUS period. The sham group (no tFUS, N = 23) is shown in grey, and the experimental group (N = 23) is shown in red. Green square denotes the tFUS stimulation. Yellow shadings denote significant differences between the tFUS and sham groups. Data are shown as Mean ± S.E.M. C-F, Phase lag index (PLI) change of EEG gamma band. Black squares indicate PLI changes involving ipsilateral and contralateral S1 regions defined in A. Blue color indicates a decrease in PLI, and orange color indicates an increase in PLI. C-D, tFUS group PLI change of EEG gamma band between tFUS (0–20 ms) and pre-tFUS (−20–0 ms) periods (C), as well as between post-tFUS (20–200 ms) and pre-tFUS (−20–0 ms) periods (D). E-F, Sham group PLI changes of EEG gamma band between tFUS (0–20 ms) and pre-tFUS (−20–0 ms) periods (E), as well as between post-tFUS (20–200 ms) and pre-tFUS (−20–0 ms) periods (F). G-J, Normalized PLI (relative to pre-tFUS period) in theta (G), alpha (H), beta (I) and gamma (J) bands between tFUS group and sham group, segregated into S1 regions (tFUS red, sham grey) and non-S1 regions (tFUS orange, sham blue). Data are shown as Mean ± S.E.M., with statistical comparisons made through two-way ANOVA and post hoc tests with Bonferroni correction for multiple comparisons (*p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001).

F. Statistical Analysis.

All statistical analyses between groups and time points were conducted using two-way ANOVA and post hoc tests with Bonferroni correction for multiple comparisons. The analysis was performed using GraphPad Prism (v10.4.2).

III. Results

A. Seizure Model Is Stable up to 4 Hours.

Fig. 2A-B contains example seizures and the corresponding increases in power used to label the seizure start/stop times. Using the seizure labels, we calculated metrics to assess the stability of the model. The Seizure Durations collected from both ipsilateral and contralateral hemispheres relative to the drug injection site were not significantly different across time or between hemispheres (Fig. 2C). Because of this consistency, all of the recorded Seizure Durations for a given hemisphere were pooled into a single histogram (Fig. 2D). The majority of recorded seizures lasted less than 2 minutes, and hemispheric distributions appear relatively similar. There were no significant differences in Seizure Count between time blocks or hemispheres; however, there was a slight decrease in number of seizures in the last two hours of the experiment (Fig. 2E). As shown in Fig. 2F, percent time in seizure between sham and post 3 for the contralateral hemisphere was significantly different, which could mean that the seizure model is no longer stable during the fifth hour of the experiment. To avoid ambiguity between model failure and therapeutic efficacy, the third block of post-stimulation was excluded from the comparison between sham and tFUS.

Fig. 2.

Seizure labeling and model characterization. A, EEG waveforms containing seizures. Top graph shows three example seizures taken from a representative baseline recording. Bottom graph shows a closeup of the middle seizure. B, Power spectrums of the same waveforms shown in panel A. Top graph contains three seizures, and bottom graph is a closeup of the second seizure in the top graph. The start (red) and stop (black) times of each seizure are labeled by vertical lines. C-F, Characterizations of seizure model from the ipsilateral (blue) and contralateral (orange) S1 relative to the 4-AP cocktail injection site. C, Average seizure duration across time for the sham condition (N = 6). Each data point represents one seizure. D, Seizure duration histogram pooling all recorded seizures regardless of the time block in which the seizure was collected (N = 6). The seizure duration distribution across time is shown in Fig. S1. E, Number of seizures counted in each time block (N = 6). Data are shown as Mean ± S.E.M. F, Percent time in seizure during each time block, calculated by summing the seizure durations from a given window and dividing by the duration of the recording. Data are shown as Mean ± S.E.M. Statistical comparisons are done using two-way ANOVA, and post hoc tests with Bonferroni correction are used for multiple comparisons (*P < 0.05, N = 6).

B. tFUS of the CC Suppresses Seizures During and After Stimulation

To quantify the effectiveness of tFUS stimulation for seizure suppression, the sham and experimental group’s seizure characteristics were compared. Fig. S3 shows an example of the seizure suppression observed in the raw EEG signal during and after stimulation. Within the tFUS group, Seizure Count relative to baseline decreased similarly in both the ipsilateral and contralateral S1. (Fig. 3A). The Seizure Counts in the ipsilateral S1 decreased from baseline by 9%, 36%, and 42% for stimulation, post 1, and post 2, respectively. The decrease in Seizure Count recorded from the contralateral S1 showed slightly higher reduction rates of 10%, 41%, and 42% for stimulation, post 1, and post 2, respectively. Both hemispheres displayed significant reductions in Seizure Count between baseline and post 1, as well as between baseline and post 2. Seizure Count within the sham group remained steady throughout, and there were significant differences between the sham and tFUS groups during post 1 and post 2.

Fig. 3.

Characterization of tFUS efficacy by means of seizure metrics. A-D, The sham group (no tFUS) is shown in grey, and the experimental group is shown in red. The left column represents the contralateral S1 metrics, and the right column represents the ipsilateral S1 metrics, both relative to the 4-AP cocktail injection site. Data are shown as Mean ± S.E.M., with statistical comparisons done using two-way ANOVA and post hoc tests with Bonferroni correction for multiple comparisons (*P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001). A, Number of seizures counted in each time block (N = 6 for sham, N = 10 for tFUS). B, Percent time in seizure during each time block, calculated by summing the seizure durations from a given time block and dividing by the duration of that same time block (N = 6 for sham, N = 10 for tFUS). C, Average inter-seizure interval in each time block calculated by averaging the interval between neighboring seizures (N = 6 for sham, N = 8 for tFUS). D, Average seizure duration in each time block calculated by averaging the durations of seizures recorded within the time block (N = 6 for sham, N = 10 for tFUS).

The Percent Time in Seizure significantly decreased by 33% in the ipsilateral S1 and by 35% in the contralateral S1 for stimulation, 55% for post 1, and 55% for post 2 in both hemispheres, respectively (Fig. 3B). Although Percent Time in Seizure for the sham condition decreased slightly across the experiment, there were no significant differences from baseline. Comparing the sham and tFUS groups, there were significant differences during the post 1 session in the contralateral S1 and during the stimulation and post 1 sessions in the ipsilateral S1. Relative to baseline, the ipsilateral S1 Inter-Seizure Interval increased by 27%, 49%, and 28% during stimulation, post 1, and post 2, respectively (Fig. 3C). The contralateral S1 Inter-Seizure Interval increased by 30%, 52%, and 31% during stimulation, post 1, and post 2, respectively. Only the post 1 session in the contralateral S1 showed significant differences relative to baseline. The Inter-Seizure Interval of the sham group remained steady across the length of the experiment, and there were significant differences between sham and tFUS during post 1 in both hemispheres. The Seizure Duration showed a steady decline across time for both the sham group and the tFUS group (Fig. 3D); however, the only significant reduction occurred from baseline to post 2 in both hemispheres of the tFUS group.

C. tFUS of the CC Modulates Brain Activity.

To explore the effect of tFUS on brain activity, we compared the PSD changes across time and between groups (Fig. 4). There were no significant changes in the theta, alpha, or gamma power ratios in either hemisphere (Fig. 4A-B, 4D). However, the beta ratio significantly decreased during tFUS and post 1 compared to baseline (Fig. 4C). Within the sham group, the power ratios showed no significant change among time blocks or between hemispheres (Fig. S4). Normalized AI of the tFUS group increased significantly within the theta and alpha bands relative to baseline, and each frequency band displayed a significant difference between the normalized AI of the tFUS group and sham group (Fig. 4E-F). Compared with the pre-4-AP baseline, CC tFUS improved interhemispheric balance (Fig. S5).

Fig. 4.

Absolute frequency power and asymmetry index (AI) comparison between tFUS and sham. A-D, The sham group (no tFUS, N = 6) is shown in grey, and the experimental group (tFUS, N = 10) is shown in red. A-D Absolute frequency power in the theta (A), alpha (B), beta (C), and gamma (D) bands for data collected from the contralateral S1 (left) and ipsilateral S1 (right) relative to the 4-AP cocktail injection site. E-H, Normalized AI (relative to the baseline recording for each condition) calculated using PSD values within the theta (E), alpha (F), beta (G), and gamma (H) bands. Data are shown as Mean ± S.E.M., with statistical comparisons made through two-way ANOVA and post hoc tests with Bonferroni correction for multiple comparisons (significant differences between groups: *P < 0.05; **P < 0.01; significant differences to baseline within a group: # P < 0.05).

D. tFUS of the CC Affects Brain Network Connectivity.

The PLI was measured to quantify connectivity, especially between the ipsilateral and contralateral hemispheres; comparisons were made between the S1 regions, as well as the non-S1 regions (Fig. 5A), to better understand how the spatial selectivity of CC tFUS (Fig. S2) affects the brain network. First, we analyzed the connectivity changes in wild-type rats. From EPs induced by CC tFUS (Fig. 5B), the PLI of the gamma band decreased during and after tFUS in most brain regions (Fig. 5C-D), whereas the gamma PLI increased slightly in the sham group (Fig. 5E-F). On average, the gamma PLI of the tFUS group decreased in both the S1 and non-S1 regions relative to the pre-tFUS window (Fig. 5J). The PLI in theta, alpha, and beta bands only decreased within the S1 regions during the tFUS conditions (Fig. S6, Fig. 5G-I). Second, we examined the connectivity changes in epileptic rats. The S1 region of the tFUS group showed decreases in PLI within theta, alpha, beta, and gamma bands during stimulation, post 1, and post 2 compared to baseline. Comparatively, the non-S1 regions did not exhibit a reduction in PLI values due to stimulation (Fig. 6). The PLI differences between wild-type and seizure rats may reflect distinct baseline network states (Fig. S7).

Fig. 6.

Functional connectivity changes in rats with seizures. A-D, Normalized PLI (relative to baseline) of sham group (no tFUS, N = 6) and experimental group (tFUS, N = 10) are compared between S1 and non-S1 regions (defined in Fig. 5A) within theta (A), alpha (B), beta (C) and gamma (D) bands. Data are shown as Mean ± S.E.M., with statistical comparisons made through two-way ANOVA and post hoc tests with Bonferroni correction for multiple comparisons (*p < 0.05, **p < 0.01 between groups).

These results indicated that tFUS of the CC is spatially selective and could significantly decrease the connectivity of the seizure onset zone and mirror S1.

IV. Discussion

In this study, we applied CC tFUS on male rats with cortical seizures induced by repeated injections of 4-AP into the right S1. Our results demonstrated that CC tFUS could significantly suppress seizures by reducing seizure occurrence, increasing the interval between seizures, and shortening seizure duration. Moreover, the suppressive effect lasted up to 2 hours following 1 hour of stimulation. CC tFUS decreased absolute beta band power, improved interhemispheric balance, and decreased the synchronization between the seizure onset zone and mirrored focus. To the best of our knowledge, this study is the first to apply tFUS to white matter in the context of epilepsy.

Our results have shown that tFUS targeting the CC can bilaterally reduce percent time in seizure by 33–35% during stimulation and 55% in the ~two hours following stimulation. Compared to invasive neuromodulation studies, the therapeutic effect of white matter tFUS is less pronounced during stimulation but lasts longer after acute stimulation and eliminates surgical risk. In a study applying invasive LFS to the CC of a 4-AP rodent model, Couturier and colleagues demonstrated a 95% reduction in percent time in seizure within the contralateral motor cortex. However, the percent time in seizure returned to pre-stimulation levels in the hour following LFS [9]. The authors conducted a similar study with seizures originating in the S1 and found that LFS of the CC reduced seizures by 65% at the focus and 97% at the mirror (contralateral) focus. There was no apparent post-stimulation effect, with seizure rates returning to pre-stimulation levels [7]. Another study applying LFS to the ventral hippocampal commissure (VHC) of a status epilepticus rodent model showed a 90% reduction in seizure frequency during two weeks of stimulation, but they also showed a 57% reduction in the two weeks following stimulation [31]. These results were not compared to a control group, so it is possible that the reduction in seizures may be due to an attenuation of the seizure model.

Researchers have attempted to explain the efficacy of tFUS for seizure suppression by analyzing electrophysiological data recorded from rodent models of epilepsy. Commonly, power changes in specific frequency bands are analyzed as evidence of altered neural excitability and network synchronization [40]. Several studies have reported that tFUS can broadly reduce the PSD of the recorded signal [6], [16] with some claiming specific frequency band decreases dependent on the stimulation parameters used [3], [12]. Some studies showed that increased PSD within the beta band might be related to cortical hyperexcitability [41], which is a predisposition of seizures [42] and could be used as a biomarker [43]. Our work showed a reduction in the absolute power of the beta band during and after CC tFUS, which may indicate a disruption in this pathological synchrony (Fig. 4).

A few studies have investigated network level changes and found that tFUS could reduce the strength of the brain network [6], [44] and modulate the nonlinear dynamics of the recorded signal in epileptic rodents [45]. Our results indicated that seizures decreased AI, reflecting a disruption of normal interhemispheric balance (Fig. S5), while CC tFUS increased AI (Fig. 4), potentially by enhancing inhibitory control or rebalancing excitatory/inhibitory activity across hemispheres. This effect may localize epileptiform activity and reduce seizure propagation [36], [46]. In our study, we observed a decrease in PLI indicative of a reduction in brain network connectivity (Fig. 5 and Fig. 6). Similarly, one study suggested that tFUS can inhibit the strength of the epileptic brain network [6], [44]. Moreover, our results showed that CC stimulation mainly inhibited the interhemispheric connection of regions that were specifically innervated by the stimulated CC fibers (Fig. 5 and Fig. S2). Because CC fibers are topographically organized, only a subset of those fibers is activated by focused stimulation [9].

Invasive LFS on white matter is thought to produce a long-lasting hyperpolarization mediated by GABA_B inhibitory postsynaptic potentials and slow afterhyperpolarization [47], whereas CC tFUS might activate axons by mechanical stimulation of elastic interface waves along the axonal membrane, leading to the generation of coupled electrical potentials in neurons [48], [49]. Although tFUS to the CC is not as effective at suppressing seizures as invasive stimulation, potentially due to their different mechanisms, it does provide a safer alternative. With an optimized parameter combination, tFUS could potentially yield similar seizure reduction rates as those observed for invasive stimulation. Furthermore, tFUS exhibits a post-stimulation effect that is not observed in acute rodent invasive studies [7], [9] and is comparable to multi-day LFS of white matter in rodents [31]. This suppressive effect likely lasts longer than the observed ~2 hours, however, our results are limited by the acute nature of the 4-AP seizure model and its attenuation with time. Future studies could apply this approach to a chronic model of epilepsy, which would help quantify the post-stimulation effects, as well as give an indication of this technique’s clinical translatability. Even though some studies have already presented the neuroprotective effect of low-intensity tFUS [21], such as inhibiting axonal injury caused by trauma [22], further in vivo and ex vivo tests need to be designed to show the safety of white matter tFUS before clinical translation. In addition, potential sex-specific differences in the effects of tFUS on epilepsy will be important to investigate in future studies, as current findings on sex differences in epilepsy incidence and in neuromodulation outcomes are inconsistent [50], [51], [52].

V. Conclusion

This work is the first to show that white matter tFUS can significantly inhibit acute seizures, exhibiting a reduction in percent time in seizure, seizure duration, and seizure count, as well as an increase in inter-seizure interval. White matter tFUS may inhibit the hyperexcitability of the epileptic brain by decreasing beta band absolute power and disrupting seizure spread by increasing the asymmetry index. Furthermore, we observed a reduction in phase lag index indicative of desynchronization of the brain network. Overall, tFUS applied to the corpus callosum is a novel non-invasive neuromodulation approach, that has the potential to treat drug-resistant epilepsy patients.

Supplementary Material

This article has supplementary downloadable material available at https://doi.org/10.1109/TNSRE.2025.3644273, provided by the authors.