Thermoacoustic and photoacoustic sensing of temperature

Thermoacoustic and photoacoustic sensing of

temperature

Manojit Pramanik

Lihong V. Wang

Washington University in St. Louis

Department of Biomedical Engineering

Optical Imaging Laboratory

Campus Box 1097

One Brookings Drive

St. Louis, Missouri 63130

Abstract.

We present a novel temperature-sensing technique using

thermoacoustic and photoacoustic measurements. This noninvasive

method has been demonstrated using a tissue phantom to have high

temporal resolution and temperature sensitivity. Because both photoa-

coustic and thermoacoustic signal amplitudes depend on the tempera-

ture of the source object, the signal amplitudes can be used to monitor

the temperature. A temperature sensitivity of

0.15 ° C

was obtained at

a temporal resolution as short as

, taking the average of 20 signals.

The deep-tissue imaging capability of this technique can potentially

lead us to

in vivo

temperature monitoring in thermal or cryogenic

applications.

2009

Society

Photo-Optical

Instrumentation

Engineers.

DOI: 10.1117/1.3247155

Keywords: thermoacoustics; photoacoustics; temperature sensing; tissue phantom;

tissue temperature

Paper 09158R received Apr. 23, 2009; revised manuscript received Jul. 7, 2009;

accepted for publication Aug. 6, 2009; published online Oct. 12, 2009.

1 Introduction

During thermotherapy or cryotherapy, it is necessary to moni-

tor the temperature distribution in the tissues for the safe

deposition of heat energy in the surrounding healthy tissue

and efficient destruction of tumor and abnormal cells. To this

end, real-time temperature monitoring with high spatial reso-

lution

1mm

and high temperature sensitivity

1°C

better

is needed.

The most accurate temperature monitoring

is by directly measuring the temperature with a thermocouple

or thermistor. However, it is invasive, hence, generally not

preferred and often not feasible. Several noninvasive tempera-

ture monitoring methods have been developed. Infrared ther-

mography is a real-time method with

0.1°C

accuracy but is

limited only to superficial temperature.

Ultrasound can be

applied for real-time temperature measurements with good

spatial resolution and high penetration depth, but the tempera-

ture sensitivity is low.

–

Magnetic resonance imaging has the

advantages of high resolution and sensitivity, but it is expen-

sive, bulky, and slow.

Therefore, an accurate, noninvasive,

real-time temperature measurement method needs to be devel-

oped.

The thermoacoustic

and photoacoustic

effects

are based on the generation of pressure waves on absorption

of microwave and light energy, respectively. A short micro-

wave and laser pulse is usually used to irradiate the tissue. If

thermal confinement and stress confinement conditions are

met, then pressure waves are generated efficiently. The pres-

sure rise of the generated acoustic wave is proportional to a

dimensionless parameter called the Grueneisen parameter, and

to the local fluence. The local fluence depends on the tissue

parameters, such as the absorption coefficient, scattering co-

efficient, and anisotropy factor, and does not change signifi-

cantly with temperature. However, the Grueneisen parameter,

which depends on the isothermal compressibility, the thermal

coefficient of volume expansion, the mass density, and the

specific heat capacity at constant volume of the tissue,

changes significantly with temperature. Thus, the generated

TA/PA signal amplitude changes with temperature. Here, we

show that by monitoring the change in the TA/PA signal am-

plitude, we were able to monitor the change in temperature of

the object.

The TA/PA technique has been widely applied in biomedi-

cal imaging applications, such as breast cancer imaging, brain

structural and functional imaging, blood-oxygenation and he-

moglobin monitoring, tumor angiogenesis, and, recently, for

molecular imaging.

–

Lately, PA sensing has also been used

to monitor tissue temperature.

–

However, TA sensing of

temperature has never been studied. These two techniques do

not interact and can be used independently. Depending on the

need, one has to choose which technique to use. The main

difference between these two techniques is the contrast

mechanism. For example, water and ion concentrations are

the main sources of contrast in TA measurements, whereas

blood and melanin are the main sources of contrast in PA

measurements. Therefore, if we need to monitor the tempera-

ture of a blood vessel, then the PA technique will be more

useful; whereas if we need to monitor the temperature of

muscles, then the TA technique will be preferred. TA/PA tem-

perature sensing is a noninvasive, real-time method. The

TA/PA technique has the ability to image deeply

up to

5cm

with high spatial resolution

scalable: millimeters to microns

We can monitor temperature with high temporal resolution

and high temperature sensitivity

scalable with temporal res-

olution:

0.015

and

0.15°C

200 s

2000 measurements

averaged

and

20 measurements averaged

resolutions,

1083-3668/2009/14

Address all correspondence to: Lihong Wang, Washington University in St.

Louis, Department of Biomedical Engineering, Optical Imaging Laboratory,

Campus Box 1097, One Brookings Drive, St. Louis, MO 63130. Tel:

314

935-

6152; Fax:

314

935-7448; E-mail: lhwang@biomed.wustl.edu

Journal of Biomedical Optics 14

, 054024

September/October 2009

Journal of Biomedical Optics

September/October 2009

Vol. 14

054024-1

Downloaded From: http://biomedicaloptics.spiedigitallibrary.org/ on 07/06/2016 Terms of Use: http://spiedigitallibrary.org/ss/TermsOfUse.aspx

respectively

. Because microwaves penetrate more deeply

into tissue than light, we can potentially monitor temperature

in vivo

for locations deep inside the body.

2 Theoretical Background

If the microwave/laser excitation is much shorter than both

the thermal diffusion

i.e., the excitation is in thermal confine-

ment

and the pressure propagation

i.e., the excitation is in

stress confinement

in a heated region, the fractional volume

expansion

/

can be expressed as

=−

where

is the isothermal compressibility,

is the thermal

coefficient of volume expansion, and

and

denote changes

in pressure

measured in Pascal

, and temperature

in Kelvin

respectively.

When the fractional change in volume is negligible under

rapid heating, the local pressure rise immediately after the

microwave/laser excitation pulse can be derived as

where

denotes mass density,

denotes specific heat capac-

ity at constant volume,

is the specific optical/microwave

absorption, and

, is the percentage of absorbed energy that

is converted to heat.

We define the Grueneisen parameter

dimensionless

where

is the velocity of sound,

denotes the specific heat

capacity at constant pressure, and

is the temperature of the

object.

Therefore,

Thus, in practice, the measured pressure signal generated due

to the microwave/laser excitation can be used to monitor the

temperature. Note that, here we always refer to the base tem-

perature of the object, not the change in temperature due to

the microwave/laser heating. The instantaneous temperature

increase in the object due to the microwave/laser pulse heat-

ing is on the order of milliKelvin and its effect on the Grue-

neisen parameter is negligible. The base temperature of the

object is a slowly varying parameter compared to the transient

temperature increase induced by a microwave/laser pulse.

3 System Description

Figure

shows the combined TA and PA system used for

sensing temperature. A similar concept of integrating light

with microwaves was used earlier for a breast cancer imaging

system.

The plastic chamber containing the sample holder

was filled with mineral oil, a nonmicrowave-absorbing mate-

rial. Moreover, because mineral oil is visibly transparent, light

absorption is negligible. Mineral oil also acts as a coupling

medium for sound propagation, and thus, mineral oil was an

ideal choice as a background medium for all our experiments.

The microwave/laser assembly was placed under the sample

holder chamber, from where it illuminated the sample by ei-

ther microwave or laser, alternately, for TA/PA sensing. The

microwave was delivered using a horn antenna, whereas the

laser was delivered by a free-space optical assembly. The

prism and ground glass of the laser illumination system were

incorporated inside the microwave horn antenna. As a result,

it was not necessary to mechanically switch between the mi-

crowave and laser sources: the switching was electronic and

instantaneous. Light was delivered through a drilled

10-mm

-diam hole in one of the narrow walls of the horn

antenna. The laser beam was broadened by a concave lens

placed outside the hole in the horn antenna, then reflected by

the prism and homogenized by the ground glass. This type of

beam expansion scheme has been used extensively

before.

The insertion of the optical devices inside the

microwave horn antenna had no significant effect on the mi-

crowave delivery.

3.1

Microwave Source

3.0-GHz

microwave source produced pulses of

0.5-

width, with a repetition rate of up to

100 Hz

. An air-filled

pyramidal horn-type antenna

WR284 horn antenna W/EEV

flange, HNL Inc.

with a rectangular opening of

7.3

10.7 cm

was used to deliver the microwaves to the

sample. The specific absorption rate of the tissue was within

the IEEE safety standards.

The horn antenna was designed to transport the transverse

electric

mode of electromagnetic

waves, so

the electric field was parallel

or nearly parallel for a horn

the surface of either narrow side

-polarized in our system

and approached zero near the inner surface of either narrow

Fig. 1

Schematic of the experimental setup. MO: mineral oil bath,

SH: sample holder vial, UST: ultrasonic transducer, GG: ground glass,

PR: prism, AN: horn antenna, CL: concave lens, TH: thermistor,

signals from a LDPE vial

i.d. 12 mm, volume 5 cc

filled with mineral

oil

and DI water

PA signals from LDPE vial filled with

ink solution

=30 cm

−1

and mineral oil at 532 nm wavelength.

Pramanik and Wang: Thermoacoustic and photoacoustic sensing of temperature

Journal of Biomedical Optics

September/October 2009

Vol. 14

054024-2

Downloaded From: http://biomedicaloptics.spiedigitallibrary.org/ on 07/06/2016 Terms of Use: http://spiedigitallibrary.org/ss/TermsOfUse.aspx

wall. By contrast, the electric field was nonzero near the sur-

face of either wide wall. Therefore, opening the light delivery

hole on the narrower side of the horn antenna

or wave guide

minimized the power leakage.

3.2

Laser Source

The light source was a Q-switched Nd:YAG laser with a rep-

etition rate of

10 Hz

, providing

6.5-ns

-wide laser pulses. The

laser system could provide

400-mJ

maximal output energy at

532-nm

wavelength. The incident laser fluence on the sample

surface was controlled to

20 mJ

/

, conforming to the

American National Standards Institute safety standards.

3.3

Detection of Ultrasound

For detecting the ultrasound signal, we used a

13-mm

-diam

active-area nonfocused transducer operating at

2.25 MHz

central frequency

ISS

2.25

0.5

COM, Krautkramer

. The

signal was first amplified by a low-noise pulse amplifier

5072PR, OlympusNDT

, then filtered electronically, and fi-

nally recorded using a digital oscilloscope

TDS640A, Tek-

tronix

. When microwaves were the illumination source, a

delay/pulse generator

SRS, DG535

triggered the microwave

pulses and synchronized the oscilloscope. By contrast, during

laser illumination, the sync out from the laser system synchro-

nized the laser pulses and the data acquisition.

3.4

Temperature Sensor

A precision thermistor for laboratory applications

sealed

PVC tip, resistance of

2252

25°C

and accuracy of

0.1°C

; ON-401-PP, Omega

was used to measure the tem-

perature of the sample. The tip of the thermistor was inserted

inside the sample to get an accurate measurement of the tem-

perature. A voltage divider circuit with a dc source

Vizatek,

MPS-6003L-1

converted the resistance change of the ther-

mistor into voltage, which was recorded using the aforemen-

tioned oscilloscope. Because the current through the ther-

mistor was very small, the self-heating was negligible.

3.5

Sample Holder Vial

Low density polyethylene

LDPE

vials with

12-mm

i.d.

and

5-cc

volume were the sample holders for the entire study.

LDPE has an acoustic impedance

=1.79

and acoustic

loss=2.5 dB

/

5 MHz

. The background material was

mineral oil, which has an impedance

=1.19

. Intensity re-

flectivity

assuming normal incidence

−

/

=0.04

. Intensity transmittivity

assuming

normal incidence

/

=0.96

96%

. The

speed of sound in

LDPE=1.95 mm

/

; thus, the wavelength

of sound in

LDPE=0.975 mm

2 MHz

. The wall thickness

of the LDPE vial was

1mm

, which was comparable to the

wavelength. Therefore, the loss of acoustic signal due to im-

pedance mismatch was relatively small

only

. More-

over, the loss of signal due to absorption inside the LDPE was

also very small,

2.5 dB

/

5 MHz

i.e., for

1mm

thick-

ness, the loss due to absorption would be

0.25 dB

/

5 MHz

2.5%

loss per mm of thickness

. Moreover, our

ultrasound transducer had a center frequency at

2.25 MHz

The lower the sound frequency is, the less the absorption is.

3 GHz

, the loss tangents of water and LDPE are 0.157

and 0.00031, respectively, and the relative dielectric constants

are

and

2.26

, respectively. The amount of power

measured in watts per cubic meter

that is absorbed is given

by,

tan

, where

is the frequency

measured

in Hertz

=8.854

−12

−1

is the relative dielectric

constant,

tan

is the loss tangent, and

is the potential gra-

dient

measured in volts per meter

. Thus,

water

LDPE

water

LDPE

tan

water

tan

LDPE

2.26

0.157

0.00031

= 17479.

Therefore, the microwave absorption of LDPE is negligible

compared to that of water. Thus, there was virtually no pos-

sibility of producing TA signals from the LDPE vial because

there was essentially no microwave absorption by the LDPE.

Figure

shows the TA signal generated from the sample

holder

LDPE vial

filled with deionized

water and with

mineral oil. The DI water vial produced a

48-mV

peak-to-

peak TA signal

red line

, while no significant TA signal was

generated from the vial filled with mineral oil

blue line

Therefore, the LDPE vial did not generate any detectable TA

signal. We also tested the PA signal generated from the LDPE

vial filled with ink solution and mineral oil, at

532 nm

wave-

length light. Black acrylic artists ink was diluted with water to

have an absorption coefficient of

30 cm

−1

532 nm

Cary 50 ultraviolet/visible spectrophotometer was used to find

the absorption coefficient of the ink solution. Figure

shows the PA signals, and the ink vial produced a

156-mV

peak-to-peak PA signal

red line

, while with mineral oil did

not produce any significant PA signal

blue line

. Therefore,

we concluded that the sample holder vial had no detectable

effect either in TA or PA signal generation and, henceforth, all

signals observed were considered to be generated from the

sample placed inside the vial.

3.6

Experimental Procedure

The sample holder vial was filled with different samples, DI

water for TA measurements, and ink solution for PA measure-

ments. Two types of experiments were done. A heated sample

was allowed to come to room temperature by natural convec-

tion, exchanging heat with the background medium

mineral

oil

. The volume of the sample was very small compared to

the mineral oil; thus, the temperature rise of the mineral oil

was neglected. The TA/PA signal was recorded with time as

the sample temperature decreased to room temperature. The

thermistor was inserted inside the sample to monitor the ac-

tual temperature. Next, cold sample was allowed to reach

room temperature by natural convection, exchanging heat

with mineral oil, and the TA/PA signal was recorded with time

as the sample temperature increased to room temperature. The

actual temperature of the sample was also monitored using a

thermistor as before. Note that, for the decreasing and increas-

ing temperature experiments, the sample holder position may

have altered slightly. The sample holder was removed, refilled

with cold/hot sample, and then placed back in the system.

Pramanik and Wang: Thermoacoustic and photoacoustic sensing of temperature

Journal of Biomedical Optics

September/October 2009

Vol. 14

054024-3

Downloaded From: http://biomedicaloptics.spiedigitallibrary.org/ on 07/06/2016 Terms of Use: http://spiedigitallibrary.org/ss/TermsOfUse.aspx

4 Results and Discussions

Figure

shows the peak-to-peak TA signal amplitude and

the actual temperature of the DI water. The TA signal de-

creased as the DI water cooled to room temperature with time.

The TA signal follows the actual temperature profile

red line

very well. Figure

plots the TA signal versus the tempera-

ture of the sample. The green line shows a linear curve fit with

of 0.95. A linear relationship between the actual tem-

perature and the TA signal was observed. Figure

shows

the TA signal increased when the cold DI water temperature

reached room temperature. The TA signal follows the actual

temperature profile

red line

very well. Figure

plots the

TA signal versus the temperature of DI water, and the green

line shows a linear curve fit with an

of 0.91. A linear

relationship between the temperature and the TA signal was

observed. For water in this temperature range, the Grueneisen

parameter is a linear function of temperature

and, there-

fore, the TA signal amplitude also varies linearly with the

temperature.

Similar experiments were done for PA measurements with

ink solution as a sample. Figure

shows the PA signal

generated from the ink solution and the actual temperature.

Once again, the PA signal decreased as the solution tempera-

ture approached room temperature, and followed the actual

temperature profile

red line

very well. Figure

plots the

PA signal versus temperature, with the green line showing a

linear curve fit with an

of 0.98. Next, a cold ink solution

was allowed to reach room temperature. The PA signal in-

creased as the temperature of the solution increased

Fig.

and followed the actual temperature profile

red line

very

well. Figure

plots the PA signal versus the temperature

with the green line showing a linear curve fit with an

0.98.

Table

summarizes the TA/PA measurements. The number

of measurements averaged was 20 for each data point, and as

the microwave/laser was operating at

10 Hz

pulse repetition

rate, the temporal resolution was

. We can see the change

in signal per degree change in temperature is slightly higher

for increasing temperature than for decreasing temperature. It

was 3.6% for increasing temperature

compared to 3.0% for

decreasing temperature

in the case of TA measurements, and

5.9% for increasing temperature

compared to 4.1% for de-

creasing temperature

in the case of PA measurements.

Figure

shows how the Grueneisen parameter of water

varies with temperature

/

in the temperature range

of interest.

It increases linearly with temperature, with a

higher slope within the range

0–20°C

than within the range

20–100°C

the slope in the range

0–20°C

is 1.48 times the

slope in the range

20–100°C

. This agrees with our observa-

tion in both TA and PA measurements for increasing and de-

creasing temperature. For the TA measurements, the slope in

the range

4–22°C

was 1.16 times the slope in the range

23–58°C

, and for the PA measurements the slope in the

range

13–22°C

was 1.44 times the slope in the range

24–46°C

. We can also see a slight difference in the signals,

depending on the direction from which the sample comes to

equilibrium. In the case of TA measurements, the equilibrium

signals were 35.1 and

28.9 mV

17.6% difference

for equi-

librium reached from cooling and warming, respectively. The

corresponding equilibrium temperatures were 22.59 and

21.66°C

. There was an almost

1°C

difference between the

two equilibrium temperatures. After accounting for this differ-

ence, we see a

14%

17.6%−

3.0%+3.6%

/

2=14.3%

signal discrepancy. In the case of PA measurements, the equi-

librium signals were 166.7 and

140.3 mV

15.8% difference

and the equilibrium temperatures were 23.57 and

21.41°C

respectively. There was an almost

2°C

temperature difference

between the two equilibrium temperatures. After accounting

for this difference, we see a

15.8%−2

4.1%

Fig. 2

TA signal and actual temperature of DI water as heated DI

water was allowed to come to room temperature,

TA signal versus

temperature, which shows almost a linear relationship. Green line is

the linear curve fitting with an

of 0.95.

TA signal and the actual

temperature of DI water as cold DI water was allowed to come to

room temperature,

TA signal versus temperature, which shows al-

most a linear relationship. Green line is the linear curve fitting with an

of 0.91.

Color online only.

Fig. 3

PA signal and actual temperature of diluted black ink solu-

tion

=30 cm

−1

as the heated sample was allowed to come to

room temperature,

PA signal versus temperature, which shows

almost a linear relationship. Green line is the linear curve fitting with

of 0.98.

PA signal and the actual temperature of ink solution

as cold solution was allowed to come to room temperature,

signal versus temperature, which shows almost a linear relationship.

Green line is the linear curve fitting with an

of 0.98.

Color online

only.

Pramanik and Wang: Thermoacoustic and photoacoustic sensing of temperature

Journal of Biomedical Optics

September/October 2009

Vol. 14

054024-4

Downloaded From: http://biomedicaloptics.spiedigitallibrary.org/ on 07/06/2016 Terms of Use: http://spiedigitallibrary.org/ss/TermsOfUse.aspx