Wireless Thermal Neuromodulator for Long-term in Vivo Cooling

Performance Assessment

A. M. Miranda

, C. Silva

, V. Silva

and P. M. Mendes

CMEMS, University of Minho, 4800-058 Guimarães, Portugal

Keywords: Implantable Device, Thermal Neuromodulator, Biomedical Wireless Device, Focal Cooling.

Abstract: Focal cooling is considered a potential solution to stop or control epileptic activity. However, despite the

available proof of concept studies, such approach requires further validation before being used in humans.

One hindering factor is the lack of suitable devices to enable large-scale validation of such methodology. This

paper presents a wireless thermal neuromodulator that can wirelessly record the rat’s brain electrical activity

and temperature. At the same time, the temperature is reduced without the need to use cumbersome liquid

pipes. The proposed device has two modules: one headstage with a cooler and sensors, and one backpack with

acquisition electronics and wireless communication capability. It is possible to record the brain temperature,

the EEG at 16 kbps, and to control the cooler’s temperature, with an autonomy of 1 day.

1 INTRODUCTION

Thermal neuromodulation has been proposed as a

solution to handle medication resistant neurological

disorders, being epilepsy one of such diseases. It has

been demonstrated that focal cooling is a potential

solution to stop epileptic activity (Fujii, 2010).

Despite such validation in controlling neuronal

activity, and before becoming an effective solution,

further tests are required to demonstrate the large

scale and long-term efficacy of focal cooling on

epilepsy control. However, the available cooling

devices are not suitable to perform the required large

scale, long-term testing in vivo. Such tests may be

performed using rats and will require the

development of a suitable device to control the rats’

brain temperature in specific spots (Fernandes, 2018).

To be placed on a rats’ brain, the cooler must satisfy

size and biocompatibility constraints (Mata, 2005).

To be transported by the rodent, the system must be

wireless and must not be oversized or overweighed.

In this paper, the design of a wireless, small, low-

power thermal neuromodulator is presented in order

to reduce the rodent’s brain temperature, while EEG,

and focus temperature is being recorded.

These authors contributed equally to this paper

2 COOLER ARCHITECTURE

The proposed cooling device uses two modules to

reduce at maximum the weight and volume on top of

the rats’ head. Placed in contact with the brain, is the

thermal neuromodulator. This component is

connected to the unit of acquisition which is inserted

in a bag on the rat’s back. Fig.1 shows the proposed

system architecture.

Figure 1: Wireless neuromodulator system overview.

The two modules were designed taking into

account their specific constraints. Their design will be

explained next.

THERMAL

ACTUATOR

ELECTRONICS

BATTERY

Miranda, A., Silva, C., Silva, V. and Mendes, P.

Wireless Thermal Neuromodulator for Long-term in Vivo Cooling Performance Assessment.

DOI: 10.5220/0006936100670072

In Proceedings of the 6th International Congress on Neurotechnology, Electronics and Informatics (NEUROTECHNIX 2018), pages 67-72

ISBN: 978-989-758-326-1

3 ACQUISITION AND CONTROL

SYSTEM

The acquisition and control system on the rats’ back

should communicate wirelessly with the outside

world using a low power solution. The proposed

device is shown below, in Fig.2.

Figure 2: Block diagram of the proposed system.

The system has two fundamental parts. The first

one, placed on the freely moving rat, and the second

one attached to the PC. The rats’ module integrates a

small and thin battery, a Bluetooth 5 (BL5) module

(ISP1507), a low-power electrophysiology signal

acquisition chip (RHD2216) and a Peltier module

used to cool down the epileptic focus.

Since it must sit on the back of a rat, this device

must be lightweight, and have a small volume, so it

will not become uncomfortable for the rodent. If so,

the animal will try to remove the device with its paws.

According to literature, a rat heavier than 250 g (an

adult rodent often is heavier than this) can carry 60 g

on its back and move freely without any

inconvenience on locomotion or motivation

(Hampson, 2009). Hence, the rat can have, positioned

on its back, a board with a size of 2.5x5 cm

without

any disturbance. Nevertheless, the size and the weight

should be reduced to the maximum possible

(Ativanichayaphong, 2008).

Our device will have near 10 g and will occupy a

volume of 4 x 4 x 0.3 cm

The second part uses also a BL5 module, which

will be connected to the PC to record and process the

data sent wirelessly from the rats’ device.

BL5 was selected, since it offers improvements in

numbers of bytes that can be sent in each connection,

allowing higher speed without compromising energy

efficiency (DiMarco, 2017). The Bluetooth 5 module

ISP1507 consumes only 7,5 mA on tx with 4 dBm or

5,3 mA with 0 dBm and 5,4 mA when working on rx.

It allows, also, for better range. BL5 grants a data

packet to be sent at a rate of 2 Mbps between 2

devices and communication in a range of 200 m

outdoors and 40m indoors (Collotta, 2017). However,

the ISP1507 reports only 100 m free open space

which is more than necessary for this application

since we only need a range of 3 to 5 meters for the

system to work correctly.

The main disadvantage of using this type of

communication standard is the difficulty to achieve

high data rate communications. However, since we

want to record an EEG at sampling rates between 250

and 2000 Hz the data rate is enough. Since we have

an 8-bit resolution ADC, we only need a data rate of

8*(sampling rate) bits per second (bps). For instance,

if we select a sampling rate of 2000 Hz, the data rate

of 16 kbps will be feasible with BL5.

The ISP1507 is from Insight Sip and it is based on

nRF52832 integrated with decoupling and loading

capacitors, 2 crystals (32 MHz and 32,768 kHz) and

a RF matching circuit and antenna. The integration of

the antenna is important, since it allows for full

system size reduction (8x8x1 mm

3.1 System Main Features

The implemented device allows to record both the

temperature and the EEG, and to control Peltier

temperature. Since the device runs on batteries, it also

sends the battery levels to recharge or to replace it.

Temperature and EEG will be described next with

more details.

3.1.1 Temperature Control

When desired, a command can be sent from the PC to

the device, to turn on the Peltier positioned on the

rat’s brain. It activates one PWM unit that digitally

encodes an analog signal level. This modulation

technique uses a square wave in which the time the

signal is on (high) or off (low) can be controlled. To

characterize the amount of time the signal is on we

can set the duty cycle to the percentage aimed. For

instance, if the duty cycle is set to 75%, then the

output from the PWM would be a square wave with

high voltage 75% of the time. If, for example, the

supply is 3 V, the resulting analog signal would be of

2.25 V (Barr, 2001). The frequency of the square

wave can also be controlled. Finally, since the output

current from the PWM generated by the chip is low

and not enough to power the peltier, it was amplified

using a Darlington pair connected between the PWM

and the Peltier. A Darlington pair acts as a single

transistor and generates a high current gain. This way,

the Peltier can be turned on and off wirelessly using

the BL5 module.

BATTERY LEVEL

EEG SIGNAL

PWM

(DUTTYCYCLE

FREQUENCY)

BATTERY

NRF52832

RHD2216

PELTIER

DARLINGTONPAIR

ELECTRODES

RAT

NEUROTECHNIX 2018 - 6th International Congress on Neurotechnology, Electronics and Informatics

3.1.2 EEG Acquisition

To detect when the rat is having a seizure, its

frequency and duration, it is necessary to record its

electroencephalogram (EEG). This is done using 3

electrodes placed on its cortical surface. These

electrodes will be connected to the electrophysiology

chip on the device. The selected chip was the

RHD2216 from Intan Technologies. It integrates 16

amplifiers, analog and digital filters and a

multiplexed analog-to-digital converter (ADC). Its

miniature size (4.8x4.1 mm

) and low-power make it

ideal for this application. It can sample 16 differential

amplified channels at a maximum data rate of

30 kSamples/s each.

This chip communicates with the BL5 module

over a digital Serial Peripheral Interface bus (SPI).

Each digital signal sent over this bus is transmitted on

a single wire. So, this communication consists of 4

standard signals. An active-low chip select (𝐶𝑆

) and

a serial data clock (SCKL) provided by the BL5

module (master device); a MOSI (master out, slave

in) data line in which a 16-bit command word flows

from the BL5 module to the Intan chip and a MISO

(master in, slave out) data line where the response

from the Intan chip (also a 16-bit word) flows to the

BL5 module.

3.2 Software Description

There are two possible network configurations in BL5

specification: connection (bidirectional

communication), and broadcast (unidirectional

communication). Since we need both BL5 modules to

communicate with each other, a connection topology

will be implemented.

In connection mode, a peripheral device sends

advertising packets. The central device receives the

advertisement and accepts the connection. Once this

is established, the peripheral stops advertising and

both devices start trading data (Collotta, 2017).

The peripheral’s data can be classified in services

and characteristics. The service is a collection of

characteristics describing a function of the peripheral.

A peripheral can have multiples, or only one service,

and a service can, also, have only one or multiple

characteristics. Each service and characteristic uses

an UUID (universally unique identifier) to identify

itself.

Only two services were created on the peripheral

device, the Biosignals service and the Battery service.

The Biosignals service has the EEG level

characteristic and the Temperature characteristic. The

Battery service has the Battery level characteristic.

The first one is responsible of sending both the data

acquired with the Intan chip and the temperature

values and receiving the configuration of the PWM.

The second one sends a notification of low battery

when its levels reach 2V.

On the central device it was created the service

PWM with the characteristic PWM values. This

characteristic is responsible for sending the frequency

and the duty cycle for the PWM. These values are set

on the PC side and the sent over BL5.

The peripheral device will send the data as a

notification which the central device will be able to

read immediately as it changes (Hortelano, 2017).

This attribute data has a Maximum Transmission Unit

of 247 bytes, i.e., for each notification, 247 bytes are

transferred from the peripheral to the central device.

After a connection between both BL5 modules is

established, the BL5 module connected to the PC

receives the data from the rat and sends it to the PC

using the universal asynchronous receiver-transmitter

(UART).

4 MICROCOOLER DESING

The previous sections describe the system’s overall

concept, as well as the control hardware. This section

will explain the cooler and heatsink design process.

To be able to perform the required thermal

neuromodulation the cooler must guarantee that the

neuronal cells temperature does not exceed 43 ºC, to

ensure cellular integrity (Yarmolenko, 2011). On the

other hand, the cold side must reach temperatures

lower than 30 ºC to suppress epileptic seizures.

Hou et al. (2011) present the project and

development of a thermal neuromodulator to epilepsy

control. In this work, the chip permits the EEG

acquisition so that the epilepsy event can be detected

and the neuromodulator can actuate and suppress it.

The chip’s dimensions are 1.4 x 0.95 mm

. So, we can

deduce the possible dimensions of an implantable

device on the brain.

So that the cooling in situ is able to interfere with

the cerebral activity a solid-state cooler using a Peltier

component (Micropelt MPC-D403) was selected. The

device is only 2x2x1 mm

in volume and allows to

reach a net cooling up to ~50 K, depending on current

and heat loading.

4.1 Heatsink and Packaging Design

It is important to refer that once the device is

implanted in the brain, it is in direct contact with the

neuronal cells. Thus, triggering an electric current in

Wireless Thermal Neuromodulator for Long-term in Vivo Cooling Performance Assessment

the Peltier results in a temperature gradient being

generated between the cold and the hot sides of the

Peltier device. Therefore, the cold side enables neural

cooling while the hot side is responsible for

dissipating the heat that is being generated to cool

down the brain.

However, it is necessary to control the

temperatures reached at the hot side of the Peltier, to

ensure the thermal modulation of the brain does not

induce irreversible damage to the cells near the hot

plate. Moreover, the continuous increase of the

temperature on the Peltier's hot side causes a

consequent increase in the temperatures reached at

the cold end after a certain time, due to the Joule

effect. So, it is necessary to include in the heat

modulating device a heatsink that needs to be in direct

contact with the hot end side of the Peltier. The reason

to choose aluminum as the heatsink was because,

despite having a lower thermal conductivity than

copper, it is a viable and a low-cost alternative in the

manufacture of heatsinks.

For heatsink design, shown in Fig.3, special

constraints were considered: the volume required to

efficiently dissipate the heat being generated; the

need to miniaturize the device dimensions without

compromising heat dissipation.

Figure 3: Model used to simulate the microcooler (Peltier

and heatsink).

Is also important to ensure the safety and viability

of the implantable device. To do so, it is important to

meet certain requirements, namely concerning its

biocompatibility. Biocompatibility is an essential

feature that ensures the device’s long-term

implantation, avoiding the induction of irreversible

damage. Therefore, the device was isolated with a

biocompatible material. Several options like flexible

polymer substrates such as polydimethylsiloxane

(PDMS), polyimide (PI) or parylene (Patil, 2016),

ensure that the device can be embedded without

compromising its structure or mode of operation.

Among the options presented, PDMS is a silicone

elastomer with excellent physicochemical properties

that make it an attractive and promising material in

the development of MEMS and numerous

components in biomedical applications. PDMS has

been traditionally used as a biomaterial in catheters

and other drainage tubes, ear and nose implants and

also as insulation material in pacemakers (Mata,

2005).

In addition, PDMS handling and manipulation is

rather easy and since it is a non-toxic and

biocompatible material it can be implanted in vivo.

Also, its elastic properties permit this polymer to

recover its initial shape after undergoing long cycles

of deformation, with the particularity of standing

strong strain deformations before breaking.

Concerning thermal properties, this material is

thermally stable, acting as a thermal insulation

(thermal conductivity of 0.2 W/mK), since it does not

allow resistive heat dissipation (Mata, 2005).

Hence, in order to simulate the PDMS embedding

device, a tri-dimensional CAD model for the PDMS

structure was planned and designed.

This structure, besides protecting the cold side of

the Peltier, which is in direct contact with brain cells,

also fills the spaces between the Peltier tellurium

bismuth pellets and covers the hot side plate of the

PDMS that is not in direct contact with heatsink.

Furthermore, since the heatsink is also coated with

PDMS, the PDMS structure should be adapted to the

dimensions and shape of the heatsink.

In the next set of tests, the PDMS thickness

relationship on the thermal performance of the device

was subsequently tested (Mata, 2005).

4.2 Temperature Control Modelling

To understand how the brain temperature changes

with the control current it was necessary to model the

cooling system behavior. Such behavior will be

highly dependent on the heatsink and its dimensions.

The temperature control modelling was studied using

COMSOL, which allows modelling all systems

previously described

A tri-dimensional block of 180x110x130 mm

was used to model the brain, since studies state that

the human brain volume ranges from 650 cm

to 1260

(Lahr, 2004). Thus, a parallelepiped with the

reported dimensions was designed and the previously

presented Peltier device was included inside,

simulating an implantable neuronal device, as can be

seen in Fig. 4.

Figure 4: Simulated brain block with implanted device.

NEUROTECHNIX 2018 - 6th International Congress on Neurotechnology, Electronics and Informatics

The Peltier device was simulated considering the

materials that compose the semiconductor pellets

(bismuth telluride Bi

), the conductive bond-pads

(copper Cu) and the materials that compose the cold

and hot sides of the plates of the device (alumina or

aluminum oxide Al

) above mentioned.

4.3 Heatsink and PDMS Cover

Performance Assessment

In this section are presented simulation results for

several heatsink dimensions and PDMS thicknesses.

Initially, it was only simulated the Peltier device

and the heatsink when implanted inside the brain. The

main goal of this study is to find the minimum

dimensions of heatsink that allow maximum heating

dissipation without the temperature in cerebral cells

being higher than the threshold of 43 °C. In order to

analyze the maximum temperature (T

max

) and the

minimum temperature (T

min

) resulting from heat

transfer between the device and the brain, different

combinations of L, W and T values (L, W, T) were

considered for the heatsink dimensions. These form

the parallelepiped seen in table 1.

The simulation time was set to 20 minutes, where

only from 10 s<t≤1200 s was applied a 50 mA

current. Furthermore, it should be noted that the

initial temperature of the device considered for all

simulations (at t=0 s) was 37 °C. The reason to do it

so was since this temperature is the basal temperature

of the human body and the device is totally integrated

in the brain tissue.

From the simulation results for the 18 different

combinations tested, the following table was

obtained.

Table 1: Results of simulations with different dimensions

of heatsink (length unit is mm).

0.5

1.5

min

(°C)

30.4

29.8

29.3

max

(°C)

44.9

44.1

43.5

min

(°C)

29.7

29.1

28.7

max

(°C)

43.9

43.3

42.8

min

(°C)

29.1

28.7

28.3

max

(°C)

43.3

42.7

42.3

min

(°C)

29.0

28.6

28.3

max

(°C)

43.1

42.6

42.2

min

(°C)

28.6

28.2

27.9

max

(°C)

42.6

42.1

41.8

min

(°C)

28.2

27.9

27.6

max

(°C)

42.1

41.7

41.4

From the previous results, one can conclude that

increasing L and W values lead to a decrease in the

min

and T

max

values. These results went according to

what was expected, since a larger heatsink area

implies a greater efficiency when heat dissipation is

occurring and, consequently, the temperatures

recorded will be lower.

Regarding the thickness relationship, it has been

observed that an increase in the thickness of the

heatsink (T), for fixed values of L and W, results in

decreasing temperature values for T

min

and T

max

, as

expected, since larger thickness values lead to a

higher volume of heat dissipation. Concluding, the

larger the heatsink (5x5x1.5 mm

), the lower the

reached temperature is.

However, since it is desirable to miniaturize the

device, the smallest heatsink should be used, while

still respecting the requirements for T

min

and T

max

temperatures that do not compromise the integrity of

neuronal cells. Since the device behavior is being

simulated, it is advisable to ensure a margin of safety

for the maximum temperature that can be reached

with respect to the critical temperature, T

max

, of 43 °C.

It was, therefore, chosen a safety margin of

approximately 1 ºC.

By observing the simulation results presented in

Table 1, it is possible to conclude that 5x5x0.5 mm

heatsink results in a T

max

of 42.1°C. Compared with

other feasible dimensions that achieved

approximately the same critical temperature, these are

the dimensions for which the volume of heat

dissipation is lower and therefore were the chosen

dimensions for the heatsink.

In Fig. 5, it is possible to observe the temperature

profile of the device being tested for the chosen

heatsink dimensions. In the coloring bar presented at

the right side of the image are the maximum and

minimum temperatures registered after the simulation

has been performed.

Figure 5: Temperature profile of the device with the

5x5x0.5 mm

heatsink.

The 5x5x0.5 mm

heatsink was tested with

different PDMS thicknesses to analyze the influence

of the referred thicknesses in achieved temperatures.

Followingly are presented the device’s temperature

Wireless Thermal Neuromodulator for Long-term in Vivo Cooling Performance Assessment

profiles for PDMS thicknesses of 20, 50 and 100 µm,

respectively.

Figure 6: Temperature profile of the device with the

5x5x0.5 mm

heatsink for PDMS thicknesses: a) 20 µm, b)

50 µm, c) 100 µm.

It was observed that increasing PDMS thickness

results in an increase T

max

, which is explained by the

thermal insulation behavior of PDMS. It was also

noted that T

min

value decreased with increasing

PDMS thickness. However, lower T

min

values was

observed specially on the cold plate of Peltier and not

on the brain. It should also be referred that with the

increasing PDMS thickness the neuronal cells cooling

region is closer to the Peltier’s cold end.

It is also noteworthy that, ideally, the thinnest

PDMS thickness that guarantees biocompatibility is

desirable.

5 CONCLUSIONS

This work presents the design and implementation of

wireless thermal neuromodulator, small and light

enough to be used for long-term in-vivo testing on

rats. The proposed device records brain temperature,

EEG, and allows to control the current that will

switch the cooling element on and off. The electronics

were fully assembled and tested, while the heatsink

and cooler were fully designed and are undergoing

testing in laboratorial conditions.

ACKNOWLEDGEMENTS

This work is supported by Foundation for Science and

Technology (FCT) project PTDC/EEI-

TEL/5250/2014, by FEDER funds through POCI-01-

145-FEDER-16695 and Projecto 3599—Promover a

Produção Científica e Desenvolvimento Tecnológico

e a Constituição de Redes Temáticas.

REFERENCES

Fujii, M., Fujioka, H., Oku, T., Tanaka, N., Imoto, H.,

Maruta, Y., Nomura, S., Kajiwara, K., Saito, T.,

Yamakawa, T. and Yamakawa, T., 2010. Application

of focal cerebral cooling for the treatment of intractable

epilepsy.Neurologia medico-chirurgica, 50(9), pp.839-

844.

Fernandes, J., Vendramini, E., Miranda, A.M., Silva, C.,

Dinis, H., Coizet, V., David, O. and Mendes, P.M.,

2018. Design and Performance Assessment of a Solid-

State Microcooler for Thermal Neuromodulation.

Micromachines, 9(2), p.47.

Mata, A., Fleischman, A.J. and Roy, S., 2005.

Characterization of polydimethylsiloxane (PDMS)

properties for biomedical micro/nanosystems.

Biomedical microdevices, 7(4), pp.281-293.

Hampson, R.E., Collins, V. and Deadwyler, S.A., 2009. A

wireless recording system that utilizes Bluetooth

technology to transmit neural activity in freely moving

animals. Journal of neuroscience methods, 182(2),

pp.195-204.

Ativanichayaphong, T., He, J.W., Hagains, C.E., Peng,

Y.B. and Chiao, J.C., 2008. A combined wireless neural

stimulating and recording system for study of pain

processing. Journal of neuroscience methods, 170(1),

pp.25-34.

Di Marco, P., Skillermark, P., Larmo, A., Arvidson, P. and

Chirikov, R., 2017. Performance Evaluation of the Data

Transfer Modes in Bluetooth 5. IEEE Communications

Standards Magazine, 1(2), pp.92-97.

Collotta, M., Pau, G., Talty, T. and Tonguz, O.K., 2017.

Bluetooth 5: a concrete step forward towards the IoT.

arXiv preprint arXiv:1711.00257.

Barr, M., 2001. Pulse width modulation. Embedded

Systems Programming, 14(10), pp.103-104.

Hortelano, D., Olivares, T., Ruiz, M.C., Garrido-Hidalgo,

C. and López, V., 2017. From sensor networks to

internet of things. Bluetooth low energy, a standard for

this evolution. Sensors, 17(2), p.372.

Yarmolenko, P.S., Moon, E.J., Landon, C., Manzoor, A.,

Hochman, D.W., Viglianti, B.L. and Dewhirst, M.W.,

2011. Thresholds for thermal damage to normal tissues:

an update. International Journal of Hyperthermia,

27(4), pp.320-343.

Hou, K.C., Chang, C.W., Chiou, J.C., Huang, Y.H. and

Shaw, F.Z., 2011. Wireless and batteryless biomedical

microsystem for neural recording and epilepsy

suppression based on brain focal cooling. IET

nanobiotechnology, 5(4), pp.143-147.

Lahr, M.M. and Foley, R., 2004. Palaeoanthropology:

Human evolution writ small. Nature, 431(7012),

p.1043.

Patil, A.C. and Thakor, N.V., 2016. Implantable

neurotechnologies: a review of micro-and

nanoelectrodes for neural recording. Medical &

biological engineering & computing, 54(1), pp.23-44.

NEUROTECHNIX 2018 - 6th International Congress on Neurotechnology, Electronics and Informatics