- Email: [email protected]
Clinical thermometry, using the 27 MHz multi-electrode current-source interstitial hyperthermia system in brain tumours
Radiotherapy and Oncology 59 (2001) 227±231
www.elsevier.com/locate/radonline
Clinical thermometry, using the 27 MHz multi-electrode current-source interstitial hyperthermia system in brain tumours Robert S.J.P. Kaatee a,*, Peter C.J.M. Nowak b, Jacoba van der Zee c, Jacob de Bree d, Bart P. Kanis a, Hans Crezee d, Peter C. Levendag b, Andries G. Visser e a
Division of Clinical Physics, Department of Radiotherapy, Daniel den Hoed Cancer Center, University Hospital Rotterdam, Groene Hilledijk 301, 3075 EA Rotterdam, The Netherlands b Department of Radiotherapy, Daniel den Hoed Cancer Center, University Hospital Rotterdam, Groene Hilledijk 301, 3075 EA Rotterdam, The Netherlands c Division of Hyperthermia, Department of Radiotherapy, Daniel den Hoed Cancer Center, University Hospital Rotterdam, Groene Hilledijk 301, 3075 EA Rotterdam, The Netherlands d Department of Radiotherapy, University Hospital Utrecht, Heidelberglaan 100, 3584 CX Utrecht, The Netherlands e Radian, Joint Centre for Radiation Oncology Arnhem Nijmegen, University of Nijmegen, PO Box 9101, 6500 HB Nijmegen, The Netherlands Received 20 September 2000; received in revised form 19 January 2001; accepted 1 February 2001
Abstract Background and purpose: In interstitial hyperthermia, temperature measurements are mainly performed inside heating applicators, and therefore, give the maximum temperatures of a rather heterogeneous temperature distribution. The problem of how to estimate lesion temperatures using the multi-electrode current-source interstitial hyperthermia (MECS-IHT) system in the brain was studied. Materials and methods: Temperatures were measured within the electrodes and in an extra catheter at the edge of a 4 £ 4 £ 4.5 cm 3 glioblastoma multiforme resection cavity. From the temperature decays during a power-off period, information was obtained about local maximum and minimum tissue temperatures. The signi®cance of these data was examined through model calculations. Results: Maximum tissue temperatures could be estimated roughly by switching off all electrodes for about 5 s. Model calculations showed that the minimum tissue temperatures near a certain afterloading catheter correspond well with the temperature of the applicator inside, about 1 min after this applicator was switched off. Conclusions: Although the electrode temperatures read during heating are not suitable to assess the temperature distribution, it is feasible to heat the brain adequately using the MECS-IHT system with extra sensors outside the electrodes and/or application of decay methods. q 2001 Elsevier Science Ireland Ltd. All rights reserved. Keywords: Interstitial hyperthermia; Thermometry; Thermocouples; Brain
1. Introduction Interstitial hyperthermia (IHT) differs from external heating techniques due to its characteristic temperature distributions with hot spots around the applicators [14]. Measurements inside applicators are useful to control the maximum tissue temperature, but care has to be taken not to overestimate the tissue temperature, due to self-heating of the applicator and/or thermometer. On the other hand, measurements in thermometry catheters, i.e. without a working applicator, can give reliable information about the local minima in the temperature distribution if the sensors are placed in the centre between the applicators or at the edge of the target volume. Knowledge about the minimum temperature in the target volume is important for good treatment control [15]. Therefore, quality assurance guide* Corresponding author.
lines [6] for temperature sensors (number, locations) should be followed. However, in practice, the number of catheters that can be implanted is limited by patient tolerance. Furthermore, the homogeneity of the temperature distribution strongly depends on the electrode density. Therefore, a compromise has to be made between heating catheters and thermometry catheters, which often leads to a shortage of information about the minimum temperatures. In some situations, minimum temperature control is possible using the applicator temperatures only, either from the ratio of applicator power and temperature rise [4] or from decay measurements after turning off power. The latter requires adequate power deposition and thermal models. The use of thermometry in applicator catheters and thermometry catheters in the clinical situation will be discussed for the 27 MHz multi-electrode current-source interstitial hyperthermia (MECS-IHT) system. Promising features of MECS-IHT regarding 3D spatial temperature control have
0167-8140/01/$ - see front matter q 2001 Elsevier Science Ireland Ltd. All rights reserved. PII: S 0167-814 0(01)00310-3
228
R.S.J.P. Kaatee et al. / Radiotherapy and Oncology 59 (2001) 227±231
already been con®rmed in model calculations [9,12,17], in muscle-equivalent phantom measurements [5,9,20] and in vivo, in rhabdomyosarcomas growing in the ¯ank of a rat [7]. The clinical application of the MECS-IHT system involves treatments of high grade (grade III/IV) gliomas. In this paper, the methods of thermometry use are presented using a large (4 cm diameter) target volume implant as an example. 2. Materials and methods 2.1. The 27 MHz MECS-IHT system The 27 MHz current-source hyperthermia system can be split into three sections, i.e. a 27 MHz heating unit, a thermometry system, and a workstation for treatment planning and treatment control. The power system has 64 coherent channels divided into two groups with a phase difference of 1808. Duty cycle power steering (from 0 to 100% of a cycle time of 200 ms, in steps of 5%) is used to maintain the strict phase relation between the channels. The electromagnetic energy is deposited in the tissue by dual-electrode applicators inserted into plastic afterloading catheters. Temperature measurements are performed with seven-point constantan±manganin thermocouple probes (ELLA-CS, Czech Republic). Fast data-acquisition (all 196 channels read within 320 ms) is feasible using a high resolution (0.0058C) thermometry system [3]. A UNIX-workstation provides the connection between thermometry, power supply and operator. A more extensive description of the multi-electrode current-source (MECS) system has been given by Lagendijk et al. [13]. 2.2. Placement of afterloading catheters, applicators and thermometers In general, target diameters in the brain are in the range of 10±50 mm. Typically, 1±10 afterloading catheters are implanted parallel at distances of about 5±10 mm apart from each other. These catheters were made of a special type of nylon with a low dielectric loss factor (tan d < 0:002), minimizing absorption of electromagnetic energy in the catheter wall. The implant discussed in this paper consisted of ten catheters implanted into a 4 £ 4 £ 4.5 cm 3 glioblastoma multiforme resection cavity. One of the catheters was used for thermometry only. In the other catheters, dual-electrode applicators were inserted, consisting of two 10 mm long electrodes with a 5 mm gap between the electrodes and a seven-point manganin±constantan thermocouple probe inside with measuring points spaced every 7.5 or 10 mm [8]. In the thermometry catheter, the temperature was measured at 14 points with a 5 mm resolution. The electrodes were placed inside the target volume to avoid hot spots in the surrounding normal tissue. All tip-electrodes are connected to heating channels with the same phase. The phase of the other electrodes is shifted 1808.
2.3. Treatment schedule and goal After surgery, the patients received 33 fractions of 1.8 Gy external radiotherapy (ERT) followed by a combination of IHT and pulsed dose rate (PDR) interstitial radiotherapy (IRT). The latter consisted of four fractions/day (with 3 h intervals) of 2 Gy/fraction and an overall dose of 24 Gy over 5 days including a weekend radiation break. On the ®rst day of the brachytherapy treatment, three IHT-sessions of about 2 h were administered between the PDR-fractions. The treatment goal was to achieve a minimum temperature of 408C in the treatment volume for as long as possible. 2.4. Treatment control In each electrode, one temperature sensor was selected as a sensor automatically controlling the power to the electrode [9]. A control cycle consists of a power-on period and a power-off period to measure temperatures. The power-off period is divided into a `recovery period', to allow the electronic disturbance of the data-acquisition equipment to disappear, and the actual `measurement period', during which temperatures are measured with a frequency of one measurement per second. Except for the recovery period, which must be at least 1 s, the length of the various periods can be chosen freely. In this case, the power-on, recovery and measurement times were 10, 2 and 3 s, respectively. During a power-off period, the temperature decreases. The temperature which is controlled by the automatic temperature control algorithm is the last temperature measured during a measurement period, which roughly represents the maximum tissue temperature just outside the afterloading catheter wall [10]. In practice, the actual value of the maximum tissue temperature was considered to be of less importance than the obtained minimum temperature in the target volume. The temperature measured in the thermometry catheter close to the edge of the target volume was taken as a ®rst estimation of the minimum temperature. The electrode target temperature was kept equal in all electrodes, and was adjusted until an edge temperature of 408C was reached. To obtain information about local minimum tissue temperatures in other regions of the target volume, occasionally, the electrodes in one of the afterloading catheters were turned off for 1±2 min. 2.5. Model calculations Model calculations were performed after the treatment of the large implant to determine the meaning of the applicator temperatures measured 1±2 min after the power was switched off in relation to local minimum tissue temperatures. The models used for the computation of the power absorption density and the resulting temperature distribution have been described by De Bree et al. [2] and Kotte et al. [11], respectively. The patient tissue geometry was obtained using CT-data and was divided into voxels with a resolution of 1 mm in
R.S.J.P. Kaatee et al. / Radiotherapy and Oncology 59 (2001) 227±231
three dimensions. Each voxel was classi®ed as either brain tissue, air or bone, depending on its Houns®eld value. No distinction could be made between grey and white matter. The tissue properties, i.e. electrical conductivity (s ), relative permittivity (1 r), density (r ), speci®c heat capacity (c), and thermal conductivity (k), used for the calculation are given in Table 1. The air was kept at a ®xed temperature of 258C. Heat transfer from skin to the air was taken into account using a heat transfer coef®cient of 8.1 W m 22 K 21. Since discrete vessel data were not available, blood ¯ow was incorporated in the model using the heat sink theory [16]. Blood ¯ow rates (wb) of 0.5 and 50 (initial value) ml/ 100 g tissue per min were chosen for bone and brain tissue, respectively. The latter was varied to match the computational results with the clinical measurements. The speci®c heat capacity coef®cient of blood (cb) was set at 3825 J kg 21 K 21. Each metal electrode and the afterloading catheter in which it was inserted were modelled together as a currentsource electrode. Since a voxel size of 1 mm is too large for accurate numerical modelling of the interaction between electrodes and the surrounding tissue, this was implemented analytically [2,18]. This was possible because the electrodes are described as geometrical objects, independent of the tissue grid resolution. The goal of the simulations was to reproduce the clinical temperature decay measurements for 1±2 min by varying the blood ¯ow rate of the brain tissue. In an iterative procedure, the electrode powers, i.e. the current injections, were modi®ed and the resulting power absorption and temperature distributions were calculated until all steady-state electrode temperatures reached a certain target temperature. Subsequently, the electrodes in one catheter were turned off, and the power absorption was calculated for this new situation. Next, the temperature decrease was calculated and compared with the clinical decay measurements. The whole simulation was repeated while the brain blood ¯ow rate was varied until a good agreement between computation and measurement was found.
3. Results In the upper panels of Fig. 1, the measured temperature decrease during multiple control cycles is shown for two situations: (1), the applicator in the central catheter #8 was turned off (left); and (2), this was done for the applicator in catheter #9, which was located at the edge opposite to the
229
edge of the thermometry catheter #1 (right). To keep the ®gures clear, only the temperatures measured in the electrodes close to the skull in catheters #8 and #9 and at the edge in the thermometry catheter are given and only the last measured temperature per control cycle is shown. Note that the electrode temperatures at t Poff 0 left and right are different. This is because the electrode target temperatures needed to reach 408C in catheter #1 were higher, i.e. 458C, at the moment that catheter #8 was turned off, than later on during the treatment when 448C was suf®cient. The temperature decay in catheter #8 demonstrates that the central region of the target volume has been at least 448C. Furthermore, the temperature in catheter #9 remains fairly stable and is still above 418C after 60 s. The measurements presented in the upper panels of Fig. 1 could be reproduced through model calculations using a blood ¯ow of 23.5 ml/100 g tissue per min. With this value, the measured temperature decays in both electrodes of the applicators in the catheters #8 and #9 were in good agreement with the calculations. This agreement was also seen in other treatments, even in very small implants (one heating catheter). This is illustrated in the lower right panel of Fig. 1, which shows the calculated temperature decay in the skull-side electrode after switching off the applicator in catheter #9. The lower left panel shows the calculated steady-state temperature distribution in a plane through the skull-side electrodes, perpendicular to the afterloading catheters with all applicators switched on. The electrode target temperature was 448C. Local cold spots between the catheters were not below 408C. Furthermore, it can be seen that all edges of the target volume, i.e. at the convex hull around and at 5 mm distance from the electrodes, were above 408C.
4. Discussion It is clear that temperature measurements in IHT require a critical interpretation. Electrode temperatures (Telectrode) are a poor measure for minimum temperature in the target volume (Tmin), and in the case of the MECS system, the maximum tissue temperature just outside the afterloading catheters (Tmax) is overestimated due to energy absorption in the applicator and in the afterloading catheter [10,19]. By measuring the temperature after switching off the power, no radio-frequency RF ®ltering of the thermocouple signals is needed and a better estimation of the maximum tissue temperature is achieved. The interval between power-off
Table 1 Tissue properties used for the model calculations a Medium
s (V 21 m 21)
1 r (..)
r (kg m 23)
c (J kg 21 K 21)
k (W m 21 K 21)
Air Brain Bone
0 0.39 0.04
1 150 9
1.3 1000 1600
0.001 3600 1400
0.004 0.53 0.65
a
Data obtained from the COMAC BME Task Group report of the European Society for Hyperthermic Oncology [1].
230
R.S.J.P. Kaatee et al. / Radiotherapy and Oncology 59 (2001) 227±231
Fig. 1. Upper panels: temperature decrease in the skull-side electrodes in catheters #8 and #9, and at the edge in the thermometry catheter #1, after one applicator was turned off for 1±2 min. (Left) The applicator in catheter #8 was turned off. (Right) The applicator in catheter #9 was turned off. Lower panels: (left), calculated steady-state temperature distribution in a plane across the skull-side electrodes perpendicular to the afterloading catheters; (right), calculated temperature decay in the skull-side electrode in afterloading catheter #9, after the electrodes in this catheter were switched off.
and measurement must be at least 1±2 s to allow the electromagnetic disturbance of the thermometry system to disappear and just long enough to let the Telectrode decrease to the value of Tmax at the moment power was switched off. Measurements in a muscle-equivalent agar phantom, using the same type of applicator and afterloading catheter as were used for IHT in the brain demonstrated that after a power-off interval of about 5 s, Telectrode is a good measure for Tmax [10]. However, it is dif®cult to determine the actual waiting time needed in clinical practice because it depends on the temperature gradient in the tissue near the electrode, and thus, on factors like the presence of neighbouring electrodes, the electrode spacing and the thermal properties of the tissue. Due to these remaining uncertainties and the steep decay of the electrode temperatures after turning the power off (typically 18C/s), electrode temperatures measured 5 s after power-off are only a rough estimate of Tmax. It is questionable whether the knowledge of maximum temperatures is really important for treatment control. It may be important that a certain minimum temperature elevation is reached in the whole target volume. Therefore, a practical approach for treatment control is to choose a somewhat arbitrary power-off time and adjust the last measured electrode temperatures until the wanted Tmin is reached, without looking at the actual electrode temperatures. However, this requires information about the actual
minimum tissue temperature. Placement of extra thermometry catheters could provide this information, but is limited by patient tolerance and the risks associated with implanting additional catheters. Under certain restrictions, it might be suf®cient to measure only the temperature at or just outside the edge of the tumour, at only one position, and use this value to set the temperature for all electrodes. This is only allowed if the measured edge temperature can be assumed to be the minimum temperature in the target volume. Therefore, the tissue must be reasonably homogeneous with respect to perfusion and electrical properties, and the electrodes must be located within the target volume. Furthermore, the distance between the edge sensor and the nearest electrode must be at least half of the largest distance between neighbouring electrodes to ensure that the local minima between the applicators are not lower than the measured edge temperature. Even if these conditions are met, it is recommended to stop the heating in one of the afterloading catheters periodically to investigate the local minimum tissue temperature in other parts of the target volume. The power-off time required to obtain reliable information about Tmin depends on the spacing between the catheters and blood perfusion. At present, estimation of the required waiting time by model calculations during the treatment is too time-consuming. A practical method to estimate Tmin is to wait until the measured temperature decay
R.S.J.P. Kaatee et al. / Radiotherapy and Oncology 59 (2001) 227±231
curves are largely levelled off. This method appears to be reliable in a homogeneous medium as is demonstrated in Fig. 1. It is obvious that for small implants, this method will underestimate Tmin. In these situations, either an extra thermometry catheter or suf®cient data of previous treatments from which electrode temperature decay information can be interpreted are required. Model calculations, which use the electrode temperatures as input data should be examined with care. Accurate modelling of self-heating of the applicators and the afterloading catheters is dif®cult. The power absorption inside the electrodes was omitted completely in the presented model calculations, and therefore, the calculated temperature inside the electrodes was underestimated. For the estimation of the average blood perfusion rate, the temperatures in the central electrode-voxel and its direct neighbour voxels were averaged. The target for this mean electrode temperature was set at the same value as the clinical electrode target temperature 4.5 s after power-off. Despite the uncertainties about the temperatures close to the electrodes, the calculated and measured long-term (60±120 s) temperature decays can be brought in close agreement with each other by varying only the blood perfusion rate in the thermal model (Fig. 1). Although the computations give good estimates of the temperature of cold spots related to electrode spacing, colder regions induced by larger vessels remain invisible. The effects of discrete vasculature on the temperature distribution can be studied with the present thermal models [11], but it is not easy to obtain the required vessel data. More information about spatial variations of blood ¯ow can be obtained if the long-term decay method is used to estimate blood perfusion rates per electrode region. Regarding the performance of the MECS-IHT system, it can be said that, using the heating technique described, it is feasible to reach minimum temperatures of 408C in brain implants, with small local hot spots of about 458C near the electrodes. Acknowledgements This study was supported by the Dutch Cancer Society and by Nucletron Corporation Veenendaal. Furthermore, the authors want to thank I.K.K. Kolkman-Deurloo for supplying the catheter co-ordinates needed for the model calculations. References [1] COMAC BME Task Group. COMAC BME Task Group report of the European Society for Hyperthermic Oncology: treatment planning and modelling in hyperthermia, 1992. [2] De Bree J, Van der Koijk JF, Lagendijk JJW. A 3D SAR model for current source interstitial hyperthermia. IEEE Trans Biomed Eng 1996;43:1038±1045.
231
[3] De Leeuw AAC, Crezee J, Lagendijk JJW. Temperature and SAR measurements in deep-body hyperthermia with thermocouple thermometry. Int J Hyperthermia 1993;9:685±697. [4] DeFord JA, Babbs CF, Patel UH, Bleyer MW, Marchosky JA, Moran CJ. Effective estimation and computer control of minimum tumour temperature during conductive interstitial hyperthermia. Int J Hyperthermia 1991;7:441±453. [5] Deurloo IKK, Visser AG, Morawska-Kaczynska M, Van Geel CAJF, Van Rhoon GC, Levendag PC. Application of a capacitive coupling interstitial hyperthermia system at 27 MHz; study of different applicator con®gurations. Phys Med Biol 1991;36:119±132. [6] Emami B, Stauffer PR, Dewhirst MW, et al. RTOG quality assurance guidelines for interstitial hyperthermia. Int J Radiat Oncol Biol Phys 1991;20:1117±1124. [7] Kaatee RSJP, Kampmeijer AG, Van Hooije CMC, et al. A 27 MHz current source interstitial hyperthermia system for small animals. Int J Hyperthermia 1995;11:785±796. [8] Kaatee RSJP, Crezee J, Kanis AP, Lagendijk JJW, Levendag PC, Visser AG. Design of applicators for a 27 MHz multi-electrode current source interstitial hyperthermia system: impedance matching and effective power. Phys Med Biol 1997;42:1087±1108. [9] Kaatee RSJP, Crezee J, Kanis AP, Lagendijk JJW, Levendag PC, Visser AG. Spatial temperature control with a 27 MHz current source interstitial hyperthermia system. Int J Radiat Oncol Biol Phys 1997;37:189±197. [10] Kaatee RSJP, Crezee J, Visser AG. Temperature measurement errors with thermocouples inside 27 MHz current source interstitial hyperthermia applicators. Phys Med Biol 1999;44:1499±1511. [11] Kotte ANTJ, De Bree J, Van der Koijk JF, Lagendijk JJW. Thermal model for hyperthermia treatment planning incorporating geometrical vessel description. In: Faulker K, Carey B, Crellin A, Harrison RM, editors. Quantitative imaging in oncology, London: British Institute of Radiology, 1996. pp. 87±89. [12] Lagendijk JJW, Van der Koijk JF, Visser AG, et al. Dose uniformity with the 27 MHz multi-electrode current source interstitial hyperthermia method. In: Bruggmoser G, Mould RF, et al., editors. Brachytherapy review. Proceedings German Brachytherapy Conference, Freiburg, November, 1994, Freiburg oncology series, monograph no. 1Freiburg: Albert-Ludwigs-University Freiburg FRG, 1994. pp. 38±44. [13] Lagendijk JJW, Visser AG, Kaatee RSJP, et al. The 27 MHz current source multi-electrode interstitial hyperthermia method. NucletronOldelft Activity Report 1995;6:83±90. [14] Levendag PC, Kaatee RSJP, Visser AG, et al. Interstitial radiation and/or interstitial hyperthermia for advanced and/or recurrent cancers in the head and neck: a pilot study. In: Seegenschmiedt MH, Sauer R, et al., editors. Interstitial and intracavitary thermoradiotherapy, Berlin: Springer±Verlag, 1993. pp. 233±239. [15] Oleson JR, Sim DA, Manning MR. Analysis of prognostic variables in hyperthermia treatment of 161 patients. Int J Radiat Oncol Biol Phys 1984;10:2231±2240. [16] Pennes HH. Analysis of tissue and arterial blood temperatures in the resting human forearm. J Appl Physiol 1948;1:93±122. [17] Van der Koijk JF, Crezee J, Van Leeuwen GMJ, Battermann JJ, Lagendijk JJW. Dose uniformity in MECS interstitial hyperthermia: the impact of longitudinal control in model anatomies. Phys Med Biol 1996;41:429±444. [18] Van der Koijk JF, De Bree J, Crezee J, Lagendijk JJW. Numerical analysis of capacitively coupled electrodes for interstitial hyperthermia. Int J Hyperthermia 1997;13:607±619. [19] Van der Koijk JF, Crezee J, Lagendijk JJW. Thermal properties of capacitively coupled electrodes in interstitial hyperthermia. Phys Med Biol 1998;43:139±153. [20] Visser AG, Deurloo IKK, Levendag PC, Ruifrok ACC, Cornet B, Van Rhoon GC. An interstitial hyperthermia system at 27 MHz. Int J Hyperthermia 1989;5:265±276.
www.elsevier.com/locate/radonline
Clinical thermometry, using the 27 MHz multi-electrode current-source interstitial hyperthermia system in brain tumours Robert S.J.P. Kaatee a,*, Peter C.J.M. Nowak b, Jacoba van der Zee c, Jacob de Bree d, Bart P. Kanis a, Hans Crezee d, Peter C. Levendag b, Andries G. Visser e a
Division of Clinical Physics, Department of Radiotherapy, Daniel den Hoed Cancer Center, University Hospital Rotterdam, Groene Hilledijk 301, 3075 EA Rotterdam, The Netherlands b Department of Radiotherapy, Daniel den Hoed Cancer Center, University Hospital Rotterdam, Groene Hilledijk 301, 3075 EA Rotterdam, The Netherlands c Division of Hyperthermia, Department of Radiotherapy, Daniel den Hoed Cancer Center, University Hospital Rotterdam, Groene Hilledijk 301, 3075 EA Rotterdam, The Netherlands d Department of Radiotherapy, University Hospital Utrecht, Heidelberglaan 100, 3584 CX Utrecht, The Netherlands e Radian, Joint Centre for Radiation Oncology Arnhem Nijmegen, University of Nijmegen, PO Box 9101, 6500 HB Nijmegen, The Netherlands Received 20 September 2000; received in revised form 19 January 2001; accepted 1 February 2001
Abstract Background and purpose: In interstitial hyperthermia, temperature measurements are mainly performed inside heating applicators, and therefore, give the maximum temperatures of a rather heterogeneous temperature distribution. The problem of how to estimate lesion temperatures using the multi-electrode current-source interstitial hyperthermia (MECS-IHT) system in the brain was studied. Materials and methods: Temperatures were measured within the electrodes and in an extra catheter at the edge of a 4 £ 4 £ 4.5 cm 3 glioblastoma multiforme resection cavity. From the temperature decays during a power-off period, information was obtained about local maximum and minimum tissue temperatures. The signi®cance of these data was examined through model calculations. Results: Maximum tissue temperatures could be estimated roughly by switching off all electrodes for about 5 s. Model calculations showed that the minimum tissue temperatures near a certain afterloading catheter correspond well with the temperature of the applicator inside, about 1 min after this applicator was switched off. Conclusions: Although the electrode temperatures read during heating are not suitable to assess the temperature distribution, it is feasible to heat the brain adequately using the MECS-IHT system with extra sensors outside the electrodes and/or application of decay methods. q 2001 Elsevier Science Ireland Ltd. All rights reserved. Keywords: Interstitial hyperthermia; Thermometry; Thermocouples; Brain
1. Introduction Interstitial hyperthermia (IHT) differs from external heating techniques due to its characteristic temperature distributions with hot spots around the applicators [14]. Measurements inside applicators are useful to control the maximum tissue temperature, but care has to be taken not to overestimate the tissue temperature, due to self-heating of the applicator and/or thermometer. On the other hand, measurements in thermometry catheters, i.e. without a working applicator, can give reliable information about the local minima in the temperature distribution if the sensors are placed in the centre between the applicators or at the edge of the target volume. Knowledge about the minimum temperature in the target volume is important for good treatment control [15]. Therefore, quality assurance guide* Corresponding author.
lines [6] for temperature sensors (number, locations) should be followed. However, in practice, the number of catheters that can be implanted is limited by patient tolerance. Furthermore, the homogeneity of the temperature distribution strongly depends on the electrode density. Therefore, a compromise has to be made between heating catheters and thermometry catheters, which often leads to a shortage of information about the minimum temperatures. In some situations, minimum temperature control is possible using the applicator temperatures only, either from the ratio of applicator power and temperature rise [4] or from decay measurements after turning off power. The latter requires adequate power deposition and thermal models. The use of thermometry in applicator catheters and thermometry catheters in the clinical situation will be discussed for the 27 MHz multi-electrode current-source interstitial hyperthermia (MECS-IHT) system. Promising features of MECS-IHT regarding 3D spatial temperature control have
0167-8140/01/$ - see front matter q 2001 Elsevier Science Ireland Ltd. All rights reserved. PII: S 0167-814 0(01)00310-3
228
R.S.J.P. Kaatee et al. / Radiotherapy and Oncology 59 (2001) 227±231
already been con®rmed in model calculations [9,12,17], in muscle-equivalent phantom measurements [5,9,20] and in vivo, in rhabdomyosarcomas growing in the ¯ank of a rat [7]. The clinical application of the MECS-IHT system involves treatments of high grade (grade III/IV) gliomas. In this paper, the methods of thermometry use are presented using a large (4 cm diameter) target volume implant as an example. 2. Materials and methods 2.1. The 27 MHz MECS-IHT system The 27 MHz current-source hyperthermia system can be split into three sections, i.e. a 27 MHz heating unit, a thermometry system, and a workstation for treatment planning and treatment control. The power system has 64 coherent channels divided into two groups with a phase difference of 1808. Duty cycle power steering (from 0 to 100% of a cycle time of 200 ms, in steps of 5%) is used to maintain the strict phase relation between the channels. The electromagnetic energy is deposited in the tissue by dual-electrode applicators inserted into plastic afterloading catheters. Temperature measurements are performed with seven-point constantan±manganin thermocouple probes (ELLA-CS, Czech Republic). Fast data-acquisition (all 196 channels read within 320 ms) is feasible using a high resolution (0.0058C) thermometry system [3]. A UNIX-workstation provides the connection between thermometry, power supply and operator. A more extensive description of the multi-electrode current-source (MECS) system has been given by Lagendijk et al. [13]. 2.2. Placement of afterloading catheters, applicators and thermometers In general, target diameters in the brain are in the range of 10±50 mm. Typically, 1±10 afterloading catheters are implanted parallel at distances of about 5±10 mm apart from each other. These catheters were made of a special type of nylon with a low dielectric loss factor (tan d < 0:002), minimizing absorption of electromagnetic energy in the catheter wall. The implant discussed in this paper consisted of ten catheters implanted into a 4 £ 4 £ 4.5 cm 3 glioblastoma multiforme resection cavity. One of the catheters was used for thermometry only. In the other catheters, dual-electrode applicators were inserted, consisting of two 10 mm long electrodes with a 5 mm gap between the electrodes and a seven-point manganin±constantan thermocouple probe inside with measuring points spaced every 7.5 or 10 mm [8]. In the thermometry catheter, the temperature was measured at 14 points with a 5 mm resolution. The electrodes were placed inside the target volume to avoid hot spots in the surrounding normal tissue. All tip-electrodes are connected to heating channels with the same phase. The phase of the other electrodes is shifted 1808.
2.3. Treatment schedule and goal After surgery, the patients received 33 fractions of 1.8 Gy external radiotherapy (ERT) followed by a combination of IHT and pulsed dose rate (PDR) interstitial radiotherapy (IRT). The latter consisted of four fractions/day (with 3 h intervals) of 2 Gy/fraction and an overall dose of 24 Gy over 5 days including a weekend radiation break. On the ®rst day of the brachytherapy treatment, three IHT-sessions of about 2 h were administered between the PDR-fractions. The treatment goal was to achieve a minimum temperature of 408C in the treatment volume for as long as possible. 2.4. Treatment control In each electrode, one temperature sensor was selected as a sensor automatically controlling the power to the electrode [9]. A control cycle consists of a power-on period and a power-off period to measure temperatures. The power-off period is divided into a `recovery period', to allow the electronic disturbance of the data-acquisition equipment to disappear, and the actual `measurement period', during which temperatures are measured with a frequency of one measurement per second. Except for the recovery period, which must be at least 1 s, the length of the various periods can be chosen freely. In this case, the power-on, recovery and measurement times were 10, 2 and 3 s, respectively. During a power-off period, the temperature decreases. The temperature which is controlled by the automatic temperature control algorithm is the last temperature measured during a measurement period, which roughly represents the maximum tissue temperature just outside the afterloading catheter wall [10]. In practice, the actual value of the maximum tissue temperature was considered to be of less importance than the obtained minimum temperature in the target volume. The temperature measured in the thermometry catheter close to the edge of the target volume was taken as a ®rst estimation of the minimum temperature. The electrode target temperature was kept equal in all electrodes, and was adjusted until an edge temperature of 408C was reached. To obtain information about local minimum tissue temperatures in other regions of the target volume, occasionally, the electrodes in one of the afterloading catheters were turned off for 1±2 min. 2.5. Model calculations Model calculations were performed after the treatment of the large implant to determine the meaning of the applicator temperatures measured 1±2 min after the power was switched off in relation to local minimum tissue temperatures. The models used for the computation of the power absorption density and the resulting temperature distribution have been described by De Bree et al. [2] and Kotte et al. [11], respectively. The patient tissue geometry was obtained using CT-data and was divided into voxels with a resolution of 1 mm in
R.S.J.P. Kaatee et al. / Radiotherapy and Oncology 59 (2001) 227±231
three dimensions. Each voxel was classi®ed as either brain tissue, air or bone, depending on its Houns®eld value. No distinction could be made between grey and white matter. The tissue properties, i.e. electrical conductivity (s ), relative permittivity (1 r), density (r ), speci®c heat capacity (c), and thermal conductivity (k), used for the calculation are given in Table 1. The air was kept at a ®xed temperature of 258C. Heat transfer from skin to the air was taken into account using a heat transfer coef®cient of 8.1 W m 22 K 21. Since discrete vessel data were not available, blood ¯ow was incorporated in the model using the heat sink theory [16]. Blood ¯ow rates (wb) of 0.5 and 50 (initial value) ml/ 100 g tissue per min were chosen for bone and brain tissue, respectively. The latter was varied to match the computational results with the clinical measurements. The speci®c heat capacity coef®cient of blood (cb) was set at 3825 J kg 21 K 21. Each metal electrode and the afterloading catheter in which it was inserted were modelled together as a currentsource electrode. Since a voxel size of 1 mm is too large for accurate numerical modelling of the interaction between electrodes and the surrounding tissue, this was implemented analytically [2,18]. This was possible because the electrodes are described as geometrical objects, independent of the tissue grid resolution. The goal of the simulations was to reproduce the clinical temperature decay measurements for 1±2 min by varying the blood ¯ow rate of the brain tissue. In an iterative procedure, the electrode powers, i.e. the current injections, were modi®ed and the resulting power absorption and temperature distributions were calculated until all steady-state electrode temperatures reached a certain target temperature. Subsequently, the electrodes in one catheter were turned off, and the power absorption was calculated for this new situation. Next, the temperature decrease was calculated and compared with the clinical decay measurements. The whole simulation was repeated while the brain blood ¯ow rate was varied until a good agreement between computation and measurement was found.
3. Results In the upper panels of Fig. 1, the measured temperature decrease during multiple control cycles is shown for two situations: (1), the applicator in the central catheter #8 was turned off (left); and (2), this was done for the applicator in catheter #9, which was located at the edge opposite to the
229
edge of the thermometry catheter #1 (right). To keep the ®gures clear, only the temperatures measured in the electrodes close to the skull in catheters #8 and #9 and at the edge in the thermometry catheter are given and only the last measured temperature per control cycle is shown. Note that the electrode temperatures at t Poff 0 left and right are different. This is because the electrode target temperatures needed to reach 408C in catheter #1 were higher, i.e. 458C, at the moment that catheter #8 was turned off, than later on during the treatment when 448C was suf®cient. The temperature decay in catheter #8 demonstrates that the central region of the target volume has been at least 448C. Furthermore, the temperature in catheter #9 remains fairly stable and is still above 418C after 60 s. The measurements presented in the upper panels of Fig. 1 could be reproduced through model calculations using a blood ¯ow of 23.5 ml/100 g tissue per min. With this value, the measured temperature decays in both electrodes of the applicators in the catheters #8 and #9 were in good agreement with the calculations. This agreement was also seen in other treatments, even in very small implants (one heating catheter). This is illustrated in the lower right panel of Fig. 1, which shows the calculated temperature decay in the skull-side electrode after switching off the applicator in catheter #9. The lower left panel shows the calculated steady-state temperature distribution in a plane through the skull-side electrodes, perpendicular to the afterloading catheters with all applicators switched on. The electrode target temperature was 448C. Local cold spots between the catheters were not below 408C. Furthermore, it can be seen that all edges of the target volume, i.e. at the convex hull around and at 5 mm distance from the electrodes, were above 408C.
4. Discussion It is clear that temperature measurements in IHT require a critical interpretation. Electrode temperatures (Telectrode) are a poor measure for minimum temperature in the target volume (Tmin), and in the case of the MECS system, the maximum tissue temperature just outside the afterloading catheters (Tmax) is overestimated due to energy absorption in the applicator and in the afterloading catheter [10,19]. By measuring the temperature after switching off the power, no radio-frequency RF ®ltering of the thermocouple signals is needed and a better estimation of the maximum tissue temperature is achieved. The interval between power-off
Table 1 Tissue properties used for the model calculations a Medium
s (V 21 m 21)
1 r (..)
r (kg m 23)
c (J kg 21 K 21)
k (W m 21 K 21)
Air Brain Bone
0 0.39 0.04
1 150 9
1.3 1000 1600
0.001 3600 1400
0.004 0.53 0.65
a
Data obtained from the COMAC BME Task Group report of the European Society for Hyperthermic Oncology [1].
230
R.S.J.P. Kaatee et al. / Radiotherapy and Oncology 59 (2001) 227±231
Fig. 1. Upper panels: temperature decrease in the skull-side electrodes in catheters #8 and #9, and at the edge in the thermometry catheter #1, after one applicator was turned off for 1±2 min. (Left) The applicator in catheter #8 was turned off. (Right) The applicator in catheter #9 was turned off. Lower panels: (left), calculated steady-state temperature distribution in a plane across the skull-side electrodes perpendicular to the afterloading catheters; (right), calculated temperature decay in the skull-side electrode in afterloading catheter #9, after the electrodes in this catheter were switched off.
and measurement must be at least 1±2 s to allow the electromagnetic disturbance of the thermometry system to disappear and just long enough to let the Telectrode decrease to the value of Tmax at the moment power was switched off. Measurements in a muscle-equivalent agar phantom, using the same type of applicator and afterloading catheter as were used for IHT in the brain demonstrated that after a power-off interval of about 5 s, Telectrode is a good measure for Tmax [10]. However, it is dif®cult to determine the actual waiting time needed in clinical practice because it depends on the temperature gradient in the tissue near the electrode, and thus, on factors like the presence of neighbouring electrodes, the electrode spacing and the thermal properties of the tissue. Due to these remaining uncertainties and the steep decay of the electrode temperatures after turning the power off (typically 18C/s), electrode temperatures measured 5 s after power-off are only a rough estimate of Tmax. It is questionable whether the knowledge of maximum temperatures is really important for treatment control. It may be important that a certain minimum temperature elevation is reached in the whole target volume. Therefore, a practical approach for treatment control is to choose a somewhat arbitrary power-off time and adjust the last measured electrode temperatures until the wanted Tmin is reached, without looking at the actual electrode temperatures. However, this requires information about the actual
minimum tissue temperature. Placement of extra thermometry catheters could provide this information, but is limited by patient tolerance and the risks associated with implanting additional catheters. Under certain restrictions, it might be suf®cient to measure only the temperature at or just outside the edge of the tumour, at only one position, and use this value to set the temperature for all electrodes. This is only allowed if the measured edge temperature can be assumed to be the minimum temperature in the target volume. Therefore, the tissue must be reasonably homogeneous with respect to perfusion and electrical properties, and the electrodes must be located within the target volume. Furthermore, the distance between the edge sensor and the nearest electrode must be at least half of the largest distance between neighbouring electrodes to ensure that the local minima between the applicators are not lower than the measured edge temperature. Even if these conditions are met, it is recommended to stop the heating in one of the afterloading catheters periodically to investigate the local minimum tissue temperature in other parts of the target volume. The power-off time required to obtain reliable information about Tmin depends on the spacing between the catheters and blood perfusion. At present, estimation of the required waiting time by model calculations during the treatment is too time-consuming. A practical method to estimate Tmin is to wait until the measured temperature decay
R.S.J.P. Kaatee et al. / Radiotherapy and Oncology 59 (2001) 227±231
curves are largely levelled off. This method appears to be reliable in a homogeneous medium as is demonstrated in Fig. 1. It is obvious that for small implants, this method will underestimate Tmin. In these situations, either an extra thermometry catheter or suf®cient data of previous treatments from which electrode temperature decay information can be interpreted are required. Model calculations, which use the electrode temperatures as input data should be examined with care. Accurate modelling of self-heating of the applicators and the afterloading catheters is dif®cult. The power absorption inside the electrodes was omitted completely in the presented model calculations, and therefore, the calculated temperature inside the electrodes was underestimated. For the estimation of the average blood perfusion rate, the temperatures in the central electrode-voxel and its direct neighbour voxels were averaged. The target for this mean electrode temperature was set at the same value as the clinical electrode target temperature 4.5 s after power-off. Despite the uncertainties about the temperatures close to the electrodes, the calculated and measured long-term (60±120 s) temperature decays can be brought in close agreement with each other by varying only the blood perfusion rate in the thermal model (Fig. 1). Although the computations give good estimates of the temperature of cold spots related to electrode spacing, colder regions induced by larger vessels remain invisible. The effects of discrete vasculature on the temperature distribution can be studied with the present thermal models [11], but it is not easy to obtain the required vessel data. More information about spatial variations of blood ¯ow can be obtained if the long-term decay method is used to estimate blood perfusion rates per electrode region. Regarding the performance of the MECS-IHT system, it can be said that, using the heating technique described, it is feasible to reach minimum temperatures of 408C in brain implants, with small local hot spots of about 458C near the electrodes. Acknowledgements This study was supported by the Dutch Cancer Society and by Nucletron Corporation Veenendaal. Furthermore, the authors want to thank I.K.K. Kolkman-Deurloo for supplying the catheter co-ordinates needed for the model calculations. References [1] COMAC BME Task Group. COMAC BME Task Group report of the European Society for Hyperthermic Oncology: treatment planning and modelling in hyperthermia, 1992. [2] De Bree J, Van der Koijk JF, Lagendijk JJW. A 3D SAR model for current source interstitial hyperthermia. IEEE Trans Biomed Eng 1996;43:1038±1045.
231
[3] De Leeuw AAC, Crezee J, Lagendijk JJW. Temperature and SAR measurements in deep-body hyperthermia with thermocouple thermometry. Int J Hyperthermia 1993;9:685±697. [4] DeFord JA, Babbs CF, Patel UH, Bleyer MW, Marchosky JA, Moran CJ. Effective estimation and computer control of minimum tumour temperature during conductive interstitial hyperthermia. Int J Hyperthermia 1991;7:441±453. [5] Deurloo IKK, Visser AG, Morawska-Kaczynska M, Van Geel CAJF, Van Rhoon GC, Levendag PC. Application of a capacitive coupling interstitial hyperthermia system at 27 MHz; study of different applicator con®gurations. Phys Med Biol 1991;36:119±132. [6] Emami B, Stauffer PR, Dewhirst MW, et al. RTOG quality assurance guidelines for interstitial hyperthermia. Int J Radiat Oncol Biol Phys 1991;20:1117±1124. [7] Kaatee RSJP, Kampmeijer AG, Van Hooije CMC, et al. A 27 MHz current source interstitial hyperthermia system for small animals. Int J Hyperthermia 1995;11:785±796. [8] Kaatee RSJP, Crezee J, Kanis AP, Lagendijk JJW, Levendag PC, Visser AG. Design of applicators for a 27 MHz multi-electrode current source interstitial hyperthermia system: impedance matching and effective power. Phys Med Biol 1997;42:1087±1108. [9] Kaatee RSJP, Crezee J, Kanis AP, Lagendijk JJW, Levendag PC, Visser AG. Spatial temperature control with a 27 MHz current source interstitial hyperthermia system. Int J Radiat Oncol Biol Phys 1997;37:189±197. [10] Kaatee RSJP, Crezee J, Visser AG. Temperature measurement errors with thermocouples inside 27 MHz current source interstitial hyperthermia applicators. Phys Med Biol 1999;44:1499±1511. [11] Kotte ANTJ, De Bree J, Van der Koijk JF, Lagendijk JJW. Thermal model for hyperthermia treatment planning incorporating geometrical vessel description. In: Faulker K, Carey B, Crellin A, Harrison RM, editors. Quantitative imaging in oncology, London: British Institute of Radiology, 1996. pp. 87±89. [12] Lagendijk JJW, Van der Koijk JF, Visser AG, et al. Dose uniformity with the 27 MHz multi-electrode current source interstitial hyperthermia method. In: Bruggmoser G, Mould RF, et al., editors. Brachytherapy review. Proceedings German Brachytherapy Conference, Freiburg, November, 1994, Freiburg oncology series, monograph no. 1Freiburg: Albert-Ludwigs-University Freiburg FRG, 1994. pp. 38±44. [13] Lagendijk JJW, Visser AG, Kaatee RSJP, et al. The 27 MHz current source multi-electrode interstitial hyperthermia method. NucletronOldelft Activity Report 1995;6:83±90. [14] Levendag PC, Kaatee RSJP, Visser AG, et al. Interstitial radiation and/or interstitial hyperthermia for advanced and/or recurrent cancers in the head and neck: a pilot study. In: Seegenschmiedt MH, Sauer R, et al., editors. Interstitial and intracavitary thermoradiotherapy, Berlin: Springer±Verlag, 1993. pp. 233±239. [15] Oleson JR, Sim DA, Manning MR. Analysis of prognostic variables in hyperthermia treatment of 161 patients. Int J Radiat Oncol Biol Phys 1984;10:2231±2240. [16] Pennes HH. Analysis of tissue and arterial blood temperatures in the resting human forearm. J Appl Physiol 1948;1:93±122. [17] Van der Koijk JF, Crezee J, Van Leeuwen GMJ, Battermann JJ, Lagendijk JJW. Dose uniformity in MECS interstitial hyperthermia: the impact of longitudinal control in model anatomies. Phys Med Biol 1996;41:429±444. [18] Van der Koijk JF, De Bree J, Crezee J, Lagendijk JJW. Numerical analysis of capacitively coupled electrodes for interstitial hyperthermia. Int J Hyperthermia 1997;13:607±619. [19] Van der Koijk JF, Crezee J, Lagendijk JJW. Thermal properties of capacitively coupled electrodes in interstitial hyperthermia. Phys Med Biol 1998;43:139±153. [20] Visser AG, Deurloo IKK, Levendag PC, Ruifrok ACC, Cornet B, Van Rhoon GC. An interstitial hyperthermia system at 27 MHz. Int J Hyperthermia 1989;5:265±276.