An Event-Driven Closed-Loop Ultrasound Stimulator Composed of a

Micro-Transducer and Multi-Site Electrodes in Vitro

Ryo Furukawa

, Shuichi Murakami

and Takashi Tateno

Bioengineering and Bioinformatics, Graduate School of Information Science and Technology, Hokkaido University,

Kita 14, Nishi 9, Kita-ku, Sapporo, Hokkaido 060-0814, Japan

Osaka Research Institute of Industrial Science and Technology, 2-7-1, Ayumino, Izumi, Osaka, 594-1157, Japan

Bioengineering and Bioinformatics, Faculty of Information Science and Technology, Hokkaido University, Kita 14,

Nishi 9, Kita-ku, Sapporo, Hokkaido 060-0814, Japan

Keywords: Brain Slice, Calcium Imaging, Closed-Loop System, Micromachined Transducer, Ultrasound Stimulation.

Abstract: Ultrasound neuromodulation, in which local and deep brain areas are stimulated, holds promise for clinical

applications. However, the mechanisms of action underlying the stimulation effects are still unknown. In vitro

experiments are helpful for investigating the stimulation mechanisms because they allow easy control of

extracellular conditions. Compared with closed-loop systems, conventional open-loop systems do not permit

monitoring of neural activity, and thus can lead to excessive neural stimulation. In this study, we developed

a piezoelectric micromachined ultrasound transducer (PMUT) combined with monitoring microelectrodes.

To examine the potential of our device as a neuromodulation tool, we measured the cellular responses to

generated ultrasound stimulation. Subsequently, we constructed a closed-loop system that combined our

PMUT with monitoring electrodes, and applied event-related ultrasound stimulation to brain slices in vitro.

We discuss future applications of a closed-loop ultrasound stimulation system.

1 INTRODUCTION

Ultrasound stimulation enables non-invasive

stimulation of local and deep areas of the brain, which

are difficult to achieve using conventional

electromagnetic stimulation methods (Tufail et al.,

2010; Wagner et al., 2007). However, the cellular

mechanisms of ultrasound-induced neural activation

are unclear. Furthermore, little is known about how

cellular responses to ultrasound stimulation influence

localized neural networks in vivo. For example,

indirect neural activity in the auditory pathway could

affect neuromodulation, complicating investigations

of the mechanism of action (Sato et al., 2018).

Most current brain stimulation strategies are

limited in terms of their ability to control patterns of

brain activity. This is because they generally involve

unidirectional stimulation processes, and thus do not

capture sufficient information about the neural

activity surrounding the stimulator. Flexible

https://orcid.org/0000-0001-8920-1025

https://orcid.org/0000-0002-8862-8446

https://orcid.org/0000-0001-9429-9880

regulation of neural activity could be achieved via

closed-loop approaches to brain stimulation,

including ultrasound stimulation. Accordingly,

researchers have prioritized the creation of

microelectrodes and small-scale stimulators within

the same device, along with the development of

bidirectional technological approaches that enable

simultaneous stimulation and recording of neural

activity.

When examining the cellular mechanisms

underlying the effects of ultrasonic stimulation, in

vitro experimental systems enable easy control of the

extracellular conditions around neurons. Lee et al.

developed a piezoelectric micromachined ultrasound

transducer (PMUT) that was helpful in elucidating the

cellular mechanisms of neuromodulation in

dissociated cultured neurons in vitro (Lee et al., 2019).

However, in their experiments, which used

dissociated cultured cells, the original neural

networks were rebuilt in networks with random

Furukawa, R., Murakami, S. and Tateno, T.

An Event-Driven Closed-Loop Ultrasound Stimulator Composed of a Micro-Transducer and Multi-Site Electrodes in Vitro.

DOI: 10.5220/0012307900003657

Paper published under CC license (CC BY-NC-ND 4.0)

In Proceedings of the 17th International Joint Conference on Biomedical Engineering Systems and Technologies (BIOSTEC 2024) - Volume 1, pages 95-102

ISBN: 978-989-758-688-0; ISSN: 2184-4305

connections. Furthermore, conventional open-loop

stimulators can induce excessive neural activity,

something which could be addressed by closed-loop

stimulation (Takeuchi & Berényi, 2020). Several

reports have described methods for closed-loop

ultrasound stimulation (Jo et al., 2022; Xie et al.,

2022). However, in these studies, the stimulator and

monitoring electrodes were not packaged together as

one instrument. Considering future applications for

chronic conditions, systems using an integrated

device are optimal.

In this study, we firstly describe the design and

fabrication of a PMUT combined with monitoring

microelectrodes. We used microelectromechanical

systems (MEMS) technology for microfabrication.

Next, we conducted intracellular calcium imaging in

vitro to examine whether the device could be used for

ultrasound neuromodulation. Finally, we constructed

a closed-loop system including our PMUT and

microelectrodes, and conducted event-related

ultrasound stimulation.

2 METHODS

2.1 Design, Microfabrication, and

Packaging

To locally stimulate a brain slice, we aimed to

develop a PMUT that satisfied three numerical

conditions: (i) a diaphragm resonant frequency of 500

kHz, (ii) a stimulation ultrasound pressure greater

than 65 kPa, and (iii) a diaphragm radius smaller than

600 µm (Lee et al., 2019; Oh et al., 2019).

Our PMUT consisted of the following five

components: a piezoelectric film, a silicon (Si) layer,

a SiO

membrane, top and bottom Pt/Ti electrodes,

and a Si supporting layer (Fig. 1). To convert electric

(voltage) signals into ultrasound pressures, we used a

thin film of a piezoelectric material as the transducer.

Prior to microfabrication, we conducted a

simulation using general-purpose physics simulation

software (COMSOL Multiphysics, Ver. 5.5,

COMSOL AB, Sweden) on a supercomputer system

(PRIMERGY CX 400/CX2550, FUJITSU, Japan) at

the Hokkaido University Computer Center. We

calculated the resonant frequency (500 kHz) and

determined the potential size range of the PMUT.

For microfabrication, the sizes of the circular

diaphragms were set according to the simulation

results. We created eight microelectrodes (200 × 200

µm

), patterned to monitor the neural activity of a

brain slice (Fig. 1A). The fabricated PMUT was

packaged with the printed circuit board (Fig. 1B,C).

2.2 Physical Properties

To characterize the resonant frequency of each

diaphragm, we applied a sinusoidal voltage input

from a multifunction generator (WF1947, NF Co.,

Japan) and measured the acoustic pressure using a

needle hydrophone (HY05N, Toray Engineering Co.,

Japan). Subsequently, we assessed the electrical chara-

cteristics of the microelectrodes via electrochemical

impedance spectroscopy. To characterize the

electrochemical properties of the microelectrodes on

the PMUT device, we used a potentiostat (Electro-

chemical Analyzer ALS720E, BAS Inc., Japan) with a

built-in frequency analyzer (Takahashi et al., 2019).

2.3 Preparation of Brain Slices

All animal experiments described below were

approved by the Institutional Animal Care and Use

Committee of Hokkaido University and carried out in

accordance with the National Institutes of Health

Guidelines for the Care and Use of Laboratory

Animals. Here, we used four C57BL/6J mice (two

male and two female mice, 7–10 weeks old, Japan

SLC Inc., Hamamatsu, Japan). Each mouse was

deeply anesthetized with isoflurane and decapitated.

The brains were rapidly removed and transferred into

ice-cold artificial cerebral spinal fluid (119 NaCl, 2.5

KCl, 2.5 CaCl

, 1.3 MgSO

, 1.0 NaH

, and 11.0

D-glucose, in mM, pH = 7.4). We prepared 400-µm

brain slices including the auditory cortex (Furukawa

et al., 2022).

2.4 Calcium Imaging

Before attempting closed-loop ultrasound stimulation,

we examined whether our PMUT could activate cells

in a brain slice. To observe cellular activities, we used

Fura-2 AM dye to image changes in the fluorescence

of intracellular Ca

concentrations. This enabled us

to avoid ultrasound vibration artifacts (Qiu et al.,

2019). Fura-2 AM is excitable at light wavelengths of

either 340 or 380 nm and emits fluorescence at 510

nm. We alternately measured fluorescence intensities

with excitation at 340 (F

340

) or 380 nm (F

380

) at a

frame rate of 50 Hz (Tateno, 2010). Then, we

estimated the ratiometric fluorescence intensity by

calculating the F

340

380

ratio. In the absence of

stimulation, we first monitored baseline changes in

transients for a period of 1 s. Subsequently, we

applied ultrasound stimulation (880 kHz, continuous

wave, 65.6 ± 1.8 kPa) for 1 s and imaged the Ca

transients, followed by monitoring of the Ca

transients during the recovery period for 1 s.

BIODEVICES 2024 - 17th International Conference on Biomedical Electronics and Devices

Figure 1: Structure and design of a PMUT device. (A) Schematic of the PMUT device under a brain slice. (a) Cross view. (b)

Front side of the cross section. (B) A packaged substrate combined with the PMUT device. (C) The front side of the PMUT

device with four diaphragms (four channels, chs. 1 to 4) and eight electrodes (1 to 8).

2.5 A Closed-Loop System Using

Ultrasound Stimulation

Figure 2A illustrates a conventional open-loop

ultrasound stimulation system, which evoked local

field potential (LFP) responses (Fig. 2C, left)

(Furukawa et al., 2022). Here, we examined a closed-

loop system with a feedback pathway from the

packaged PMUT (Fig. 2B). In the closed-loop

system, spontaneous brain slice activity was

monitored while ultrasound stimulation was applied

at the time point(s) when the activity exceeded a set

threshold (Fig. 2C, right). We set the threshold at 150

µV. To monitor LFPs, we used a MED amplifier

(MED-A64HS1, MED-A64MD1A, Alpha MED

Scientific, Japan) from a previous MEA-based

An Event-Driven Closed-Loop Ultrasound Stimulator Composed of a Micro-Transducer and Multi-Site Electrodes in Vitro

Figure 2: Schematic of ultrasound stimulation systems and related signals. (A) A block diagram of a conventional open-loop

ultrasound stimulation system. (B) Our constructed closed-loop system with the packaged PMUT. (C) Schematic of activity

detected via the extracellular voltage of a brain slice. The event detector sends trigger signals to the multifunction generator

immediately after detecting LFP signals that negatively cross a threshold level. The left panel shows a typical evoked LFP

response to ultrasound stimulation.

recording system (Furukawa et al., 2022). The signals

were recorded with a sampling rate of 20 kHz, and

subjected to filtering within a frequency range from 1

Hz to 10 kHz. We employed a multifunctional I/O

data acquisition system (USB-6343, National

Instruments, USA) for real-time signal control, which

acquired neural signals from the MED amplifier and

fed a trigger signal to a multifunction generator when

a threshold-crossing event was detected. We used a

customized Python program (Python Ver. 3.11.1) for

real-time processing. To detect neural activities, we

used electrode 7 for monitoring. We applied

ultrasound stimulation (540 kHz, continuous wave,

95.5 ± 1.4 kPa) from a diaphragm (ch 4, which was

close to electrode 7). For a single trial, the

measurement period was 1 min.

BIODEVICES 2024 - 17th International Conference on Biomedical Electronics and Devices

3 RESULTS

3.1 Microfabricated PMUT

According to the result obtained from our numerical

simulation, we determined the optimal thickness of

the material layers (PZT, 100 µm; Si, 15 µm; SiO

, 1

µm) and the radius (580 µm). We estimated that the

diaphragm would oscillate with a resonant frequency

of 540 kHz.

Our acoustic pressure measurement indicated

forced diaphragm oscillations driven by voltage

signals with a wide range of frequencies including

540 kHz, at which the oscillating amplitude had a

positive peak (Fig. 3A).

When we increased the amplitude of the input

voltage signal at a fixed frequency of 540 kHz, the

output acoustic pressure was monotonically increased

(Fig. 3B). The measured electrode impedance on the

PMUT device was 23.4 ± 3.7 kΩ at 1 kHz (eight

electrodes). Thus, the fabricated low impedance micro-

electrodes were able to monitor LFPs from brain slices.

3.2 Calcium Imaging

We next examined the possibility that our PMUT

could activate cells. Figure 3C shows an example of

imaging (indicator, Fura-2 AM), with the

averaged waveforms of the fluorescence intensity

ratio obtained via ratiometric imaging (19 cells from

Figure 3: Measured acoustic characteristics of the fabricated PMUT and cellular activation via ultrasound stimulation. (A)

Measured acoustic pressure vs. driving voltage signals with different sinusoidal frequencies (input voltage, 10 V). (B)

Acoustic pressure with different input voltages under a fixed frequency of 540 kHz. The channel numbers correspond to those

shown in Fig. 1D. (C) PMUT-driven average transients of ratiometric imaging (black) data related to intracellular Ca

concentrations for all analysed cells. The edges of the grey band represent the standard error of the mean at each time point

(n = 19 cells from five slices). Grey bar indicates the duration of the ultrasound stimulation (US).

An Event-Driven Closed-Loop Ultrasound Stimulator Composed of a Micro-Transducer and Multi-Site Electrodes in Vitro

Figure 4: Closed-loop system with real-time detection of spontaneous activity in a brain slice. (A) A typical example of a

spontaneous LFP waveform from a brain slice. The dashed red line indicates the threshold. (B) A typical example of

spatiotemporal LFP recorded at eight microelectrode sites on the PMUT. The horizontal blue bars represent the timings of

ultrasound stimulation (US; duration of each stimulation pulse, 500 ms). The blue circle represents the stimulation site; the

schematic diagram shows only the relative positions of the recording electrodes and the stimulating diaphragms, and the

actual distances between them differ. Electrode 7 was used for detection of spontaneous activity. In the bottom panel, the

timings of threshold crossing are illustrated via vertical red lines. The time point of 0 ms is expressed with arrows in the top

panel. (C) Schematic diagram of event detection and triggering methods. The current triggering method (left) had no

refractory (silent) period(s), whereas our future triggering method will include this feature (right). Vertical bars represent

event timings, and horizontal red lines indicate the threshold level for event detection. In the current system, the stimuli can

overlap according to the timings of events occurring within 500 ms. In our future system, the silent period(s) will ensure that

the US do not temporally overlap.

five slices). The intracellular Ca

concentration of

the cells increased in response to ultrasound

stimulation generated by the PMUT.

3.3 Event-Driven Ultrasound

Stimulation

After confirming that our PMUT could modulate

cellular activity, we next attempted to create a closed-

loop system using ultrasound stimulation. A typical

BIODEVICES 2024 - 17th International Conference on Biomedical Electronics and Devices

100

example of LFP recording of a brain slice is

illustrated in Fig. 4A. The negative peaks of rapid

voltage changes in LFPs occurred randomly, and the

peak events were detected as negative crossings

below the threshold (−150 µV). The inter-event

intervals were 15.5 ± 1.1 s for the seven channels for

one brain slice.

When each event was detected (Fig. 4B),

ultrasound stimulation was automatically applied to the

brain slice from the PMUT. The stimulation was driven

by an ultrasound signal generated by a functional

generator through a power amplifier (Fig. 2B).

In our closed-loop system (Fig. 2B), peak-event

detection was continuously functioning immediately

after the onset of ultrasound stimulation and during

each stimulus (500 ms). As a result, repetitive stimuli

were triggered without a silent period (Fig. 4B).

Therefore, a longer stimulation duration (> 500 ms)

could occur if the extracellular voltage continued to

exceed the threshold level (Fig. 4C).

For example, Fig. 4B shows how the LFP

responses rapidly decreased after the stimulation

onset and slowly returned to individual baselines at

four chs. (i.e., chs. 1, 2, 6, and 8), which were close

to the ultrasound stimulation site (diaphragm ch. 4 in

the transducer array). Since the LFP at the detection

site (i.e., electrode ch. 7) exceeded the threshold

voltage for a relatively longer period, trigger signals

at multiple timings (four times in Fig. 4B, bottom)

were sent to the multifunction generator, resulting in

a stimulation duration over 500 ms (c.f., Fig. 4C, left).

4 DISCUSSION

In this study, we developed a MEMS-based PMUT

with monitoring microelectrodes for event detection

(i.e., rapid changes in LFP magnitude). We

successfully microfabricated the PMUT with four

circular diaphragms for ultrasound stimulation and

eight microelectrodes for monitoring LFP-peak

events. To demonstrate that our device could perform

ultrasound neuromodulation, we conducted

intracellular calcium imaging. An influx of Ca

into

cells during ultrasound stimulation was successfully

observed in acute brain slices. These intracellular

transients suggest that our PMUT has potential

for ultrasound neuromodulation applications.

Subsequently, we constructed a closed-loop

system that included the PMUT as part of the

ultrasound stimulator. To the best of our knowledge,

this is the first attempt to combine a PMUT with

electrodes that monitor cellular activity as an

integrated device in a closed-loop system.

In our current system, ultrasound stimulation was

automatically applied to the target when the detected

signals were larger than the voltage threshold.

Therefore, it was unclear whether the detected signals

were truly attributable to spontaneous brain slice

activity or were the result of electrical noise. We are

planning to utilize more robust detection techniques

to detect neural activity in our future work.

Upon improvements to the real-time detector in

our system, our device could be applied to the

detection of abnormal neural activity such as seizure-

like activity (Berényi et al., 2012; Ranjandish &

Schmid, 2020). In our next detector model, we are

planning to include a silent period after each detected

event as a triggering rule (Fig. 4C, right). This

triggering rule could limit excessive stimulation.

Moreover, we plan to test this device in in vivo animal

experiments via chronic ultrasound stimulation.

ACKNOWLEDGEMENTS

R.F. was supported by Grant-in-Aid for JSPS Fellows

[grant number JP23KJ0047]. T.T. was supported by

the Murata Science Foundation, the Suzuken

Memorial Foundation, the Nakatani Foundation for

Advancement of Measuring Technologies in

Biomedical Engineering, a Grant-in-Aid for

Exploratory Research [grant number 18K19794], and

a Grant-in-Aid for Scientific Research (B) [grant

number 19H04178] (Japan).

REFERENCES

Berényi, A., Belluscio, M., Mao, D., & Buzsáki, G. (2012).

Closed-Loop Control of Epilepsy by Transcranial

Electrical Stimulation. Science, 337(6095), 735–737.

https://doi.org/10.1126/science.1223154

Furukawa, R., Kaneta, H., & Tateno, T. (2022). A

Multielectrode Array-Based Recording System for

Analyzing Ultrasound-Driven Neural Responses in

Brain Slices in vitro. Frontiers in Neuroscience,

16(February), 1–19. https://doi.org/10.3389/fnins.20

22.824142

Jo, Y., Lee, S. M., Jung, T., Park, G., Lee, C., Im, G. H.,

Lee, S., Park, J. S., Oh, C., Kook, G., Kim, H., Kim, S.,

Lee, B. C., Suh, G. S. B., Kim, S. G., Kim, J., & Lee,

H. J. (2022). General‐Purpose Ultrasound

Neuromodulation System for Chronic, Closed‐Loop

Preclinical Studies in Freely Behaving Rodents.

Advanced Science, 9(34). https://doi.org/10.1002/

ADVS.202202345

Lee, J., Ko, K., Shin, H., Oh, S.-J., Lee, C. J., Chou, N.,

Choi, N., Tack Oh, M., Chul Lee, B., Chan Jun, S., &

An Event-Driven Closed-Loop Ultrasound Stimulator Composed of a Micro-Transducer and Multi-Site Electrodes in Vitro

101

Cho, I.-J. (2019). A MEMS ultrasound stimulation

system for modulation of neural circuits with high

spatial resolution in vitro. Microsystems &

Nanoengineering, 5(1), Article 1. https://doi.org/

10.1038/s41378-019-0070-5

Oh, S. J., Lee, J. M., Kim, H. B., Lee, J., Han, S., Bae, J.

Y., Hong, G. S., Koh, W., Kwon, J., Hwang, E. S., Woo,

D. H., Youn, I., Cho, I. J., Bae, Y. C., Lee, S., Shim, J.

W., Park, J. H., & Lee, C. J. (2019). Ultrasonic

Neuromodulation via Astrocytic TRPA1. Current

Biology, 29(20), 3386-3401.e8. https://doi.org/

10.1016/j.cub.2019.08.021

Qiu, Z., Guo, J., Kala, S., Zhu, J., Xian, Q., Qiu, W., Li, G.,

Zhu, T., Meng, L., Zhang, R., Chan, H. C., Zheng, H.,

& Sun, L. (2019). The Mechanosensitive Ion Channel

Piezo1 Significantly Mediates In Vitro Ultrasonic

Stimulation of Neurons. iScience, 21, 448–457.

https://doi.org/10.1016/j.isci.2019.10.037

Ranjandish, R., & Schmid, A. (2020). A Review of

Microelectronic Systems and Circuit Techniques for

Electrical Neural Recording Aimed at Closed-Loop

Epilepsy Control. Sensors, 20(19), Article 19.

https://doi.org/10.3390/s20195716

Sato, T., Shapiro, M. G., & Tsao, D. Y. (2018). Ultrasonic

Neuromodulation Causes Widespread Cortical

Activation via an Indirect Auditory Mechanism.

Neuron, 98(5), 1031-1041.e5. https://doi.org/

10.1016/j.neuron.2018.05.009

Takahashi, S., Muramatsu, S., Nishikawa, J., Satoh, K.,

Murakami, S., & Tateno, T. (2019). Laminar responses

in the auditory cortex using a multielectrode array

substrate for simultaneous stimulation and recording.

IEEJ Transactions on Electrical and Electronic

Engineering, 14(2), 303–311. https://doi.org/10.1002

/tee.22810

Takeuchi, Y., & Berényi, A. (2020). Oscillotherapeutics –

Time-targeted interventions in epilepsy and beyond.

Neuroscience Research, 152, 87–107. https://doi.org/

10.1016/j.neures.2020.01.002

Tateno, T. (2010). A small-conductance Ca2+-dependent

K+ current regulates dopamine neuron activity: A

combined approach of dynamic current clamping and

intracellular imaging of calcium signals. NeuroReport,

21(10), 667–674. https://doi.org/10.1097/WNR.0b013e

32833add56

Tufail, Y., Matyushov, A., Baldwin, N., Tauchmann, M. L.,

Georges, J., Yoshihiro, A., Tillery, S. I. H., & Tyler, W.

J. (2010). Transcranial Pulsed Ultrasound Stimulates

Intact Brain Circuits. Neuron, 66(5), 681–694.

https://doi.org/10.1016/j.neuron.2010.05.008

Wagner, T., Valero-Cabre, A., & Pascual-Leone, A. (2007).

Noninvasive human brain stimulation. Annual Review

of Biomedical Engineering,

9, 527–565. https://doi.org/

10.1146/annurev.bioeng.9.061206.133100

Xie, Z., Yan, J., Dong, S., Ji, H., & Yuan, Y. (2022). Phase-

locked closed-loop ultrasound stimulation modulates

theta and gamma rhythms in the mouse hippocampus.

Frontiers in Neuroscience, 16(September), 1–12.

https://doi.org/10.3389/fnins.2022.994570

BIODEVICES 2024 - 17th International Conference on Biomedical Electronics and Devices

102