TISSUE HEATING DUE TO ENDOCARDIAL LEADS DURING

MRI SCANS

Numerical Models and Experimental Validation

Eugenio Mattei, Giovanni Calcagnini, Michele Triventi, Federica Censi, Pietro Bartolini

Italian National Institute of Health, Department of Technology and Health, Rome, Italy

Veronica Piacentini and Stefano Pisa

Department of Electronic Engineering, Sapienza University of Rome, Rome, Italy

Keywords: Endocardial Leads, Magnetic Resonance Imaging, Heating, Numerical Models.

Abstract: Magnetic Resonance Imaging (MRI) tissue heating due to implanted leads represents a major concern for

the safety of patients bearing metallic devices. In this work temperature measurements were performed to

validate the typical solutions adapted as endocardial lead models, that are a thin bare metal wire and an

insulated one. Both experimental and numerical analysis was performed in the frequency range between

10kHz and 128MHz ( frequencies of the gradient and RF fields of MRI systems). We found that the bare

wire is not a reliable model to study the RF heating locally induced at the lead tip. At low frequencies

(<1MHz), the PM lead can be properly modelled as an insulated thin metal wire, providing that the actual

resistivity of the lead is also modelled. As frequency increases, such a model becomes less accurate and

different solutions must be adopted.

1 INTRODUCTION

In the last years the number of patients with

implantable devices, such as pacemakers (PM),

cardioverter/defibrillators (ICD), deep brain

stimulators (DBS) has notably increased. These

patients are today considered strongly

contraindicated to undergo magnetic resonance

imaging (MRI) examinations. This is mainly due to

the hazards that the radiofrequency (RF) field and

the gradient field may represent for the implanted

patient. The energy of the MRI field can couple into

conductive leads, like that used for implantable

devices, in two major ways: 1) the conductive lead

acts as an antenna capable of receiving and

supporting the field, and this mechanism can create

the resonant waves (Pisa et al, 2008); 2) the implant

acts as an electrical ‘short circuit’ to the electrical

potentials induced within the body by the RF field

(Konings et al, 2000). Each of these effects may

create at the lead-tissue interface high electrical

current density which, combined with the high

electrical resistance of most of the biological tissues,

can cause local resistive heating. This power

deposition in tissues, that results in an increased

local absorption rate (SAR) and consequently in

temperature growing, is potentially harmful to the

patient; an excessive temperature growth may bring

living tissues to necrosis and to death. In addition,

the induced currents that flow from the implant into

the body may cause unwanted stimulation of

excitable tissues, such as muscles or nerves. In this

field, in-vivo studies are difficult to carry out for

practical and ethical reasons, whereas computational

techniques represent a favourable choice for testing

the safety of implantable devices within an MRI

environment. However, numerical studies often

implies some simplifications of the actual physical

model that is studied, which may affect the validity

and generality of the results. For this reason,

numerical studies always need to be firmly

anchored in reality and have to be validated by

experimental measurements. The aim of this paper is

the validation, through experimental measurements

of the induced heating, of the model of an

endocardial lead. The analysis was performed in the

RF range between 10kHz and 500kHz (frequencies

of the MRI gradient fields), between 32MHz and

378

Mattei E., Calcagnini G., Triventi M., Censi F., Bartolini P., Piacentini V. and Pisa S..

TISSUE HEATING DUE TO ENDOCARDIAL LEADS DURING MRI SCANS - Numerical Models and Experimental Validation.

DOI: 10.5220/0003163503780381

In Proceedings of the International Conference on Biomedical Electronics and Devices (BIODEVICES-2011), pages 378-381

ISBN: 978-989-8425-37-9

 2011 SCITEPRESS (Science and Technology Publications, Lda.)

128MHz (frequencies of the MRI RF fields). The

most commonly used lead models, as thin bare metal

wire or thin insulated metal wire, are first

considered. In addition, since in clinical practice

endocardial leads may not always have a negligible

resistive impedance, a new model that takes into

account a finite electrical conductivity of the lead is

proposed.

2 METHODS AND MATERIALS

The RF behaviour of an endocardial lead was first

validated in terms of induced heating at the tip

versus the two typical models used to represents

leads in numerical studies: a thin bare metal wire

and an thin insulated metal wire. The same structure

has been considered both numerically and

experimentally. This provided us with a direct mean

for validating the results of the numerical analysis,

thus reducing the number of variables typically

involved in other validation procedures. Due to the

wide frequency range covered by the analysis

(10kHz-128MHz), two experimental set-ups were

designed and then reproduced as numerical models:

a low-frequency (LF) set-up and a high-frequency

(HF) set-up.

2.1 Experimental Setups

2.1.1 LF setup (10kHz, 100kHz, 500kHz)

The lead was placed inside a 20cm x 14cm x 30cm

PVC box filled with gelled saline material (HEC

2%, NaCl 0.16%) that mimics the electrical and the

thermal properties of human tissues at the

considered frequency. The measured conductivity

for the gel was 0.2 S/m. A density of 1006 kg/m

, a

thermal conductivity equal to 0.2 W/(m·K) and a

specific heat of 4178 J/(kg·K) have been provided

by the HEC manufacturer.

The current was injected directly into the lead

by applying a voltage gap between one end of the

lead and a metallic electrode placed on a wall of the

PVC box. The temperature increase caused by the

current at the lead tip-gel interface was measured by

using a fluoroptic® thermometer (Luxtron model

3100) with SMM probes placed in transversal

contact with the lead tip (Mattei et al, 2007). These

plastic fiber probes (1mm diameter) have an

accuracy of 0.1°C and operated at 8 samples per

second. Temperature measurements were performed

at the tip of a PM unipolar lead (Hepta 2, Sorin

Biomedica) and at the tip of the two simplified lead

models: a 20-cm long thin bare metal wire

(radius=1 mm) and the same wire but with an

insulation sheath. A constant current of 40 mA rms

was injected in the leads at the different frequencies

of interest (10kHz, 100kHz, 500kHz), for a period

of 300s. The amplitude of the current as well as the

voltage at the gap between the lead and the

electrode on the box was measured in real-time

during the experiments.

2.1.2 HF setup (32 MHz, 64MHz, 128MHz)

For the higher frequencies we designed an

experimental set-up that allowed us to measure the

RF current all along the lead. The three leads were

placed inside a PVC tube (radius=2cm;

height=30cm) filled with the same gel previously

described, but with a conductivity of 0.6 S/m. The

RF signal, was applied between the lead and an

electrode placed on one end of the PVC tube, at the

frequency of 32MHz, 64MHz, 128MHz. The current

flowing inside the lead and the gel was measured by

using a clamp current probe (BCP512, A.H. System,

CA). This set-up allows to test the lead in a

condition similar to the implant, but since the current

probe is not submersible, we were forced to measure

the total current crossing a section containing the

lead and the surrounding gel. The temperature

increase was measured with the amplitude of the RF

signal set to have always the same current value of

40mA rms at the lead tip. In addition, we measured

the current amplitude in seven positions along the

tube (at 0cm, 5cm, 10cm, 15cm, 20cm, 25cm - lead

tip, 30cm), so to have an estimation of the current

amplitude distribution at the frequencies of 1MHz,

32MHz, 64MHz, 128MHz. Finally, temperature VS

current curves were calculated for the PM lead and

the two simplified models, in order to relate the

temperature increase measured at the led tip to the

amount of current flowing into the tissue.

2.2 Numerical Models

2.2.1 LF setup (10kHz, 100kHz, 500kHz)

Numerical simulations have been performed using a

commercial software, CST studio 2008, based on the

Finite Integral Technique. The experimental set-ups

were faithfully reproduced in the numerical

environment: the bare wire was modelled as a metal

thin conductor (radius=1mm), while the insulated

wire was modelled as the same wire with a rubber

insulation (thickness=1mm). In order to define a

new model that could better reproduce the behaviour

of an actual endocardial lead, we measured the

TISSUE HEATING DUE TO ENDOCARDIAL LEADS DURING MRI SCANS - Numerical Models and Experimental

Validation

379

electrical parameters of the PM lead with an LCR

meter and we found an impedance of about 50 Ohm.

This impedance is likely to be localized at the lead

tip, that is generally made of conductive ceramic

materials. For this reason, the insulated wire was

modified by adding at its end a cylindrical tip with

radius and height both of 1mm and with electric

conductivity σ=6.37 S/m. It results in a resistance of

about 50 Ohm. The same voltage measured in the

experimental tests was applied in the model, to

ensure the comparability between experimental and

numerical data.

2.2.2 HF set-up (32MHz, 64MHz, 128MHz)

As for the LF measurements, also the HF

experimental set-up was faithfully reproduced in the

CST environment. In this case, the excitation was

applied in terms of current injected into the leads.

The amplitude of the current was chosen to match

the experimental value measured at one end of the

lead when the current at the tip was 40 mA rms.

3 RESULTS

Both in experimental and simulated measurements,

the bare wire showed temperature increases at its tip

comparable with the resolution of the Luxtron

thermometer (0.1°C). The temperature increments

obtained from experimental and simulated data are

reported in Figure 1, both for LF and HF. For each

frequency, two couples of bars are reported: the first

one compares the heating at the tip of the insulated

wire, (experimental versus numerical); the second

one compares the temperature increase measured at

the tip of the PM lead (experimental) to that of the

50 Ohm-tip insulated wire (numerical). Figure 2

shows the current distributions measured along the

PVC tube used for HF experiments. Up to 32MHz,

no significant changes in the current amplitude were

observed from the beginning of the tube to its end.

At higher frequency, a downward trend is noticed,

so that the current at the tip is much smaller than the

one applied at the other end of the lead. At 128

MHz, the minimum for the current amplitude was

observed at 25 cm from the tube end (position of the

lead tip), afterwards the current start increasing. The

same behaviour was observed both for the PM lead

and the insulated wire, even if the decrease in

current amplitude as frequency increases was more

evident for the former than the latter. In Figure 3 the

current VS temperature curves are reported. The

current amplitude reported on the X-axis is

measured at the lead tip. As expected from the

relation between SAR and E field, a quadratic trend

was found (R

>0.95).

Figure 1: Comparison between experimental an numerical

temperature increases measured at the tip of the insulated

metal wire and of the PM lead. The numerical model of

the PM lead refers to the 50Ohms insulated wire.

Figure 2: Current amplitude distribution along the PCV

tube used for HF temperature measurements. Left panel:

insulated metal wire. Right panel: PM lead.

Figure 3: Current VS Temperature curves for the insulated

metal wire (left panel) and the PM lead (right panel).

Current values were measured inside the PVC tube, at the

section corresponding to the lead tip.

4 DISCUSSION

The bare wire represents the simplest way to model

an endocardial lead. However, our data show that it

cannot be considered a reliable solution to

investigate the RF induced heating in human tissues.

The absence of an insulation sheath produces a more

uniform power deposition along the wire, so that the

local heating at the lead tip becomes not significant.

A thin metal wire insulated by a rubber sheath is the

typical solution adopted both in experimental and

numerical studies dealing with RF induced heating

BIODEVICES 2011 - International Conference on Biomedical Electronics and Devices

380

on implanted leads (Bassen et al, 2007). We

compared this model with an actual PM lead. First

of all, we compared the temperature increase

measured at the tip of an insulated metal wire with

its equivalent numerical model: very good

agreement between simulated and experimental

results was found, for all the frequencies of interest

(10kHz-128MHz). In the LF range, temperature

increases measured at the tip of the PM lead are

much higher than for the insulated wire (more than

three times higher), for the same excitation

conditions. Numerical simulations show that an

improvement of the lead model can be achieved

considering the actual resistivity of the lead: when

modelling a wire with a finite-conductivity tip, the

gap from the real PM lead becomes significantly

lower (<28% respect to the experimental value). The

lead impedance is a characteristic that can sensibly

vary from lead to lead, mainly depending on the

properties of the tip material, When it is made of

ceramic components (such us pyrolitic carbon), the

resistivity is close to 50 Ohm, whereas, in case of

metallic materials, it goes down to few Ohm or even

less. In any case the lead manufacturers should

indicate the value of the impedance in the technical

notes of their products; thus, it should be an easy

element to take into account. At higher frequencies

(32MHz, 64MHz, 128MHz), marked differences

between the PM lead and its numerical model are

observed, even when modelling the finite-

conductivity tip wire. In particular, at 128MHz, the

temperature increase at the PM lead tip, which up to

64MHz is always higher than the insulated wire,

becomes lower. It suggests that at higher frequencies

other mechanisms play an important role in the heat

generation process, that the simple model of an

insulated wire cannot take into account. To better

understand this aspect, we measured the current

peak distribution along the lead. The higher is the

frequency, the closer is the length of the lead to the

theoretical resonance value inside the gel. At

128MHz the theoretical resonance length is about

23cm for a dipole inside a dielectric mean with a

permittivity of about 25. In such conditions, a kind

of resonant wave can be supposed inside the lead. A

resonance phenomenon in various kinds of linear

structures (e.g. catheters used in interventional

radiology) has been hypothesized also by other

groups (Nitz et al, 2001). Heating induced by the

same current measured at the section of the tip

shows opposite trend with increasing frequency in

the insulated wire and in the PM lead. This is a

further confirmation that, at high frequencies, the

simplified model of the insulated wire moves far

from the actual behaviour of the PM lead. Thus, in

order to obtain reliable results new and more

realistic models must be developed. For example, as

already proposed by some groups (Helfer et al,

2006, Neufeld et al, 2009), a model able to

reproduce the actual inductance of the lead may

represent a substantial improvement of the analysis.

5 CONCLUSIONS

Our study reveals that at frequency below 1 MHz,

the RF–induced heating on endocardial leads can be

properly evaluated by modelling the lead as an

insulated wire. It is however necessary to take into

account the resistivity of the lead, in particular the

resistivity of the tip. At higher frequencies this

model becomes less reliable and new solutions

which consider also other aspects, such as the lead

inductance, must be adopted.

REFERENCES

Bassen H., Kainz W., Mendoza G., Kellom T. MRI-

induced heating of selected thin wire metallic

implants-- laboratory and computational studies--

findings and new questions raised. Minim Invasive

Ther Allied Technol. 2006;15(2):76-84.

Helfer J. L., Gray R.W., MacDonald S.G., Bibens T.W..

Can pacemakers, neurostimulators, leads, or guide

wires be MRI safe? Technological concerns and

possible resolutions. Minim Invasive Ther Allied

Technol. 2006;15(2):114-20.

Konings M. K., Bartels L. W., Smits H. F., Bakker C. J..

Heating around intravascular guidewires by resonating

RF waves. J Magn Reson Imaging. 2000 Jul;12(1):79-

85.

Mattei E., Triventi M., Calcagnini G., Censi F., Kainz W.,

Bassen H. I., Bartolini P. Temperature and SAR

measurement errors in the evaluation of metallic linear

structures heating during MRI using fluoroptic®

probes. Physics in Medicine and Biology, 2007 Mar

21;52(6):1633-46.

Neufeld E, Kühn S., Szekely G., Kuster N. Measurement,

simulation and uncertainty assessment of implant

heating during MRI. Phys Med Biol. 2009 Jul

7;54(13):4151-69.

Nitz W. R., Oppelt A., Renz W., Manke C., Lenhart M.,

Link J. On the Heating of linear conductive structure

as Guide Wires and Catheters in Interventional MRI,

Journal of Magnetic Resonance Imaging 2001, 13

105-114.

Pisa S., Calcagnini G., Cavagnaro M., Piuzzi E., Mattei E.,

and Bernardi P, "A study of the interaction between

implanted pacemakers and the radio-frequency field

produced by magnetic resonance imaging apparatus",

IEEE Trans. Electromag. Compat., vol. 50, no. 1, pp.

35-42, February 2008.

TISSUE HEATING DUE TO ENDOCARDIAL LEADS DURING MRI SCANS - Numerical Models and Experimental

Validation

381