Personal tools
You are here: Home Research site Dissertations 2001 Jeroen van de Kamer

2001 Jeroen van de Kamer

High-Resolution Regional Hyperthermia Treatment Planning

Hyperthermia is the induced temperature rise in tumours with the objective to enhance the cell-killing effect of radio- and/or chemotherapy. This temperature rise can, for example, be realized by the absorption of electromagnetic radiation in tissue. In regional hyperthermia this principle is used to heat deep seated, large tumours. In a Dutch randomized trial it has been shown that regional hyperthermia combined with radiotherapy can be successful for tumours in the pelvic region, especially for patients with cervical cancer. For regional hyperthermia it is difficult to elevate tumours to the desired temperature range of 42-44°C due to the occurrence of systemic stress and local pain sensations caused by high temperatures (hot spots). To get more insight in hyperthermia treatments and the aforementioned problems we have developed a hyperthermia treatment planning (HTP) system.

The planning system for regional hyperthermia is based on a finite-difference time-domain computer model to compute the electric field (E-field) and absorbed power (SAR) distribution within a patient. The dielectric patient model that is required for these computations is obtained by segmenting a CT data set into fat, muscle and bone and using literature-based values of the dielectric properties for those tissue types. Because the SAR computation on CT-resolution (1 X 1 X 5 mm^3) would require 310 days on a standard personal computer, this high-resolution is down-scaled to a lower resolution of 1 cm^3 (computation time 2.8 hours). Since the values of the dielectric properties for the different tissue types are not accurately known, we investigated the effect of erroneous values on the SAR distribution. It appeared that an error of 50% in the dielectric values resulted in an error of only 20% in the SAR distribution.

The use of the HTP system reveals the close relation between SAR distribution and patient anatomy, stressing the need for high-resolution computations. To obtain high-resolution SAR distributions a method called quasistatic zooming has been developed: From the cm-resolution E-field distribution the high-resolution potential distribution on the surface of a small zoom volume is calculated. With the CT-resolution dielectric geometry of the patient and this potential distribution as boundary conditions, the high-resolution SAR distribution is computed using a quasistatic SAR model. Repeating this for several zoom volumes gives the high-resolution SAR distribution for an arbitrary volume of interest. With quasistatic zooming it takes 4.3 days to compute CT-resolution SAR distributions within a complete patient geometry. Because SAR maxima are related to hot spots it is very important that such maxima are correctly predicted by the HTP system. SAR maxima that were predicted by the low-resolution SAR model were not present in the zoomed- nor high-resolution distributions and, even worse, the low-resolution SAR model failed to predict SAR maxima that were predicted by both quasistatic zooming and high-resolution SAR modelling. This leads to the conclusion that high-resolution computations are necessary for reliable HTP.

The quasistatic zooming technique is now an integral part of our HTP system and is used to study the hot-spot phenomenon in detail.