The optimization of magnetic resonance temperature imaging sequences

1995 IEEE-EMBC a d CMBEC Theme 2 Imaging

The Optimization

of

Magnetic Resonance

Temperature Imaging Sequences

Pin-Hen Ho, Jyh-Horng Chen

Department of Electrical Engineering,

National

Taiwan

University,

Taipei,Taiwan,

R.O.C.

Purpose

In this paper, we design the optimal gradient echo pulse parameters to achieve maximal temperature sensitivity of MR signal based on measured T1

,

T2 and proton density data at different temperatures for various tissues in vitro.

Key word:

MR,

T1, T2 proton density.

1.

Introduction

MR temDerature imaging has shown its feasibility in

lab

1

-hyperthermia temperature monitoring for optimal therapy : : : : I

effects. It is unique in that non-invasive deep tissue temperature distributions could be possible with properly designed MR sequences. Its weakness is in its sensitivity to detect minute temperature chanpes in a reasonable duration of time for medical diagnosis. In

this

paper, we design the MR pulse sequences to optimize the temperature sensitivity with the T1, T2, proton density measured at different temperatures.

2. Material and method

Bxperiments was conducted in a OE SIGNA 1.5 tesla MR Imager. Different tissues, including brain, muscle and fat was wrapped by a heating water bag of which water temperature was controlled by a temperature servo. Heated water is circulating in tubes and was pumped during the experiments. RTD PTlOOO sensors and Yokogawa 7563 electroNc thermometer was used with a multiplexer for real-tixned and multiechanneled temperature monitoring. Acquired data is then

transmitted into

PC

for further analysis.

TI,T2 and PD was measured using two different TRs, double Spin echo imaging sequences with TES as 3 h s and 60ms for muscle, 4Oms and 8Oms for fat and brain. In order to obtain optimal T1 values, TRs are set to be 3sec and

0.8T1

according to different tissues. We then plot these measured T1, T2 and

PD for all these tissues as a function of temperature and fit them to a linear equation of temperature for

further

sequence optimization.

GE and SE siganl are calculated as a function of temperature and TR, TE. Sensitivity of these signal to temperature(dS/dT)

Table.1 *TR/TE=(3000/2O)ms for muscle, (3000/40)ms for,fat, (3000/40)m for brain.

The errorbar in Fig(1) is derived from noise in image calculated in error propagation algorithm to T1 * T2 and PD.

T ~optimization E ~ 1 0 , T2(T), PD(T) could be fitted f i O m ~ i g , l : Ti(T)=M1.42+12J3.T T,(T) -58.67-0594.T PD(T) = 12293

-

0.0069. T

T,(T)

= 12556

+

5.914 T

T,(T)

= 33.46 + 0394.

T

for muscle for fat can be derived and expressed as a fimction of

TR

TE and

temperature T. By setting T equal to 37.5 oc

,

which is about the temperature of human body, one can find the optimal

M R

imaging parameter TR, TE with computer program for different tissues.

3. Result

TI, T2,PD &tu

parameters are as follows

:

The relationship of various tissues's

0-7803-2475-7197 5 10.00 0 1 997 IEEE

P D Q = 12553

-

0.0075 T

T

I

(T) = 806.1 1

+

6.48 *

T

T, ('T) 1 96.07

-

0538 * T for brain

PD(T) = 1.1 192

-

0.0032

-

T

One can optimize the pulse sequence with the relationship above and calculate the signal for spin echo and gradient echo

as a fuunction of T, TE and TR. To find the optimal TR/TE, we derive dS/dT=F(T,TR,TE) and adjust TR, TE to make IdS / d q

largest at the temperature of 37.5 "C

.

As shown in Fig.2, the results are listed in Table.2, 3, and 4. Fat

14

10

Fig.2-1 Contour plot of muscle, fat and brain of optimization for spin echo

which is time-consuming in clinical use. To include the time- effectiveness and related change in signal-to noise, we optimize (dS/dT)/sqrt(time) instead of dS/dT to have a compromise

Table.3. dS/dT for spin echo sequence with time consideration

muscle fat brain

GE(TR/TE/Angle) 18/8/16 8/3/19 31/4/19

dS/dT -2.87 -3.63 -1.47

Table.4. dS/dT for gradient echo sequence with time consideration

All these sensitivity has been checked right by MR experiment.

This

justfj the effectiveness of this optimizatiuon procedure.

4. Discussion

MR Signal Sensitivity vs.

TI,

T2, PD sensitivity

To increase temperature sensitivity is our main purpose, From table l., one can expect with the combined temperature sensitivity of all three parameters T1, T2 and PD for muscle and brain, MR signal sensitivity will be higher. One typical value has been shown in the table. However, for fat tissue, T1 and T2 sensitivity to temperature will cancel out in MR signal. This implies that MR signal strength of muscle and brain are more sensitive to temperature than anyone of the three parameters, while fat is more sensitive to

T1

or T1-weighted imaging as the temperature image choice.

Time Considerations From table 4, we can extimate the imaging time to have a image of 1 O C sensitivity

.

In order to

distiguish the signal difference fiom noise, one has to consider on the value AS I N

.

For minimal AS I N to conceive the

Im la) za Eo 00 am OD a, w

Fig.2-2 Contour plot of muscle, fat and brain of optimization for spin echo with time consideration.

Signal change, Rose suggests values higher than 3. In

this situation, it takes about 29sec for muscle, 7sec for fat and 142sec for brain which make the noise of the image one-third of the signal difference per centigrade. The ROI is about 2cm square. To increase the sensitivity by 2 one has to pay four times time in hyperthermia, this is all righrt, but it is far from

I 18

%"

the diagnostic requirements to have real-timed temperature monitoring for better spatial resolution.

5. Conclusion

We have measure T1, T2, PD sensitivity to temuerature and m 22 f O 5 /, I8 15 10 14 10 2p M 40 50 5 1 0 20 30 U ) TR TR I S b YI 40 M

A

TR

Fig.2-3 Contour plot of muscle, fat and brain of optimization

optimize MR signal TE/TR parameters -for MR-temperature

imaging. To even faster the process, we are now working on

M R temperature optimization using phase imaging with the same process. With the higher inherebt sensitivity of chemical shift parameter, we expect to get better MR temperature imaging sensitivity in a shorter duration of time.

6. Reference

:

1. J. Syha, C. Hillenbrand, etal. "Temperature depending changes of in vivo TI relaxation times in the rat brain." 2nd

Meeting, SMR, I994

2. A.L. Alexander, A.F. Gniitro, etal. "Optimization of gradient- Echo Pulse Sequences for Dynamic Imaging of Hyperthermia." 2nd Meeting, SMR, 1994