The use of a semi-customized phantom for verification of conformal plans

Medical Dosimetry, Vol. 24, No. 3, pp. 183–188, 1999 Copyright © 1999 American Association of Medical Dosimetrists Printed in the USA. All rights reserved 0958-3947/99/$–see front matter

PII S0958-3947(99)00016-3

THE USE OF A SEMI-CUSTOMIZED PHANTOM FOR VERIFICATION OF CONFORMAL PLANS J. Y. GIRAUD, PH.D., C. COHARD, M.S., A. DUSSERRE, PH.D., P. VASSAL, PH.D., M. BOLLA, M.D., J. TROCCAZ, PH.D., P. CINQUIN, M.D., PH.D., Y. MENGUY, PH.D., S. GRAND, M.D., F. ESTEVE, M.D., PH.D., and J. F. LEBAS, M.D., PH.D. CHU de Grenoble, Service d’Oncologie-Radiothe´rapie, BP 217, Grenoble, France; TIMC-IMAG, Grenoble, France; and Service d’IRM, Grenoble, France (Received 2 June 1999; Accepted 7 June 1999)

Abstract—We have developed a technique for inverse treatment planning of prostate therapy designed to improve the degree of conformation between the dose distribution and the target volume. We compared the inverse plan with a “standard” four-field box technique as well as a four-field technique using oblique fields (“cross technique”). We validated the dosimetry of the inverse plan using Fricke gel solution in phantom specifically designed for this purpose. The phantom is a Plexiglas tank with a cross section, which approximates the dimensions of the pelvis. Anatomical data from computed tomography (CT) images of a patient were used to simulate organs in our phantom. This allows us to calculate dose distributions with the external geometry of the phantom and internal anatomy of the patient. Dose–volume histograms (DVHs) for the three different plans were calculated. The phantom containing the Fricke gel was irradiated according to the inverse plan. Magnetic resonance (MR) images was used to determine the dose distribution delivered to the phantom. We observe, on DVHs, that the inverse plan significantly reduces the dose to the rectum and the bladder but slightly increases the inhomogeneity inside the target volume. Correlation is good between isodoses on MR images and calculated isodoses. We conclude that inverse planning software can greatly improve the conformal degree of treatment to the prostate. This technique could be applied to other complex anatomic sites at which dose to organs at risk is a limiting factor and increased dose to the target volume is indicated. Our phantom and the Fricke gel solution are convenient to carry out validation of conformal treatments. © 1999 American Assocation of Medical Dosimetrists. Key Words: Phantom, Fricke gel, Radiation therapy, Optimization.

treatment. The aim of this paper is to validate the result of the inverse optimized calculation. For this purpose we have compared the optimized plan with a standard four field box technique as well as a different four field technique using opposed oblique (“cross technique”). We have developed a specific phantom for validation of the dosimetry of the inverse-treatment plan.

INTRODUCTION The process of treatment planning in Radiotherapy can follow a direct or indirect approach. Direct planning is typically an iterative process of treatment fields definition, dose calculation and plan evaluation. The evaluation of the plan can have two results. If the plan is acceptable to the clinician, the patient can be treated as defined by the plan. If the plan is not satisfactory, the planning process must be repeated to satisfy all the requirements of the treatment. By contrast, the inverse planning technique first defines the requirements of the treatment then an optimal plan is determined that satisfies all the requirements. Research about inverse planning has been conducted at our facility to improve the conformal degree of our treatments. We have chosen to examine prostate therapy because current treatment methods are limited by dose delivered to the critical organs. Our approach is based on quadratic and geometric optimization of the inverse problem. This method of optimization was implemented and tested for prostate

MATERIALS AND METHODS Materials Inverse planning. The inverse-planning software was developed by Menguy et al.1,2 An isocentric technique is required, and doses are normalized to the dose at the isocenter. Dose calculations are carried out using the Clarkson algorithm. The constraints of the treatment for optimization are as follows: ● ● ●

Reprint requests to: J. Y. Giraud, CHU de Grenoble, Service d’Oncologie-Radiothe´rapie, BP 217, 38043 Grenoble Cedex 9, France. E-mail: [email protected].

minimal dose to the target volume; homogeneity in the target volume; and maximum dose to the critical organs.

The optimization starts with placement of a fixed number of rectangular beams using a single isocenter. 183

184

Medical Dosimetry

Volume 24, Number 3, 1999

Fig. 1. Phantom designed to receive the Fricke-infused solution.

The isocenter is placed at the center of the smallest sphere that can contain the target volume. Sixteen coplanar treatment fields are radially distributed around the isocenter at intervals of 22.5°. The width and the length of each beam are defined by direct projection of the target. The optimization is a determination of the ideal weighting of each beam such that the dose requirements are satisfied. This procedure inevitably results in the suppression of some of the beams entirely (optimal weighting is zero). Conventional techniques. Conventional treatment usually consists of four fields. There are two common four-field arrangements: the four-field “box” technique and the “cross” technique. The “box” technique consists of anteroposterior and posteroanterior fields and two lateral fields. The “cross” technique is a 45° of rotation of the box technique. The four treatment fields are now right anterior, left anterior, right posterior and left posterior oblique, respectively. Pelvic phantom and Fricke-infused solution. We have designed a pelvic phantom (Fig. 1) to receive 10 L of a Fricke-infused solution. Its wall consists of a 5-mm thickness of Plexiglas. The Fricke-infused solution3– 6 is

made of 10⫺3 mol/L of Fe(NH4)2(SO4)3, 50 ⫻ 10⫺3 mol/L of H2SO4, 10 g/L of agarose and 10⫺3 mol/L of NaCl. It is read with a Phillips Gyroscan (Best, Neederland) 1.5T magnetic resonance imaging (MRI) system. The sequence used permits a density proportional to T1 value for each pixel of the image to be obtained.7 The thickness of each image is 1 cm. The temperature of the gel was monitored during the irradiation and the MRI acquisition at 22 ⫾ 2°C. We irradiated the Fricke-infused solution according to the optimized plan. The treatment was delivered with 25 MV photons from a Siemens (Concord, California) KD2 linac. The TPR2010 of the beam is 0.79. Patient and phantom data. The treatment planning was carried out using the external dimensions of the phantom and the internal anatomy of a patient as determined by a computed tomography (CT) scan. The CT data consisted of 100 consecutive CT slices of 2-mm thickness. The isocenter was placed at the center of the CT data. Methods Patient and phantom data were transferred by network to our treatment-planning computer. Prostate, blad-

Table 1. Description of the optimized treatment plan Jaws (cm) Field Number 1 2 3 4 5 6 7

SSD

Gantry Angle (Degrees)

Collimator Angle (Degrees)

A1

A2

B1

B2

Weight at Isocenter (%)

83.0 84.1 90.6 90.3 85.7 82.3 83.3

67.5 90 135 225 247.5 270 315

2 12 84 4 86 78 6

5.6 6.1 4.1 5.6 6.2 6.5 4.1

5.1 5.2 5.4 5.0 3.2 2.8 5.2

3.3 2.9 5.1 5.6 5.1 5.1 5.6

6.3 6.5 5.6 3.6 5.8 6.2 4.1

14.6 24.8 7.0 31.1 3.8 11.5 7.2

SSD, source–surface distance.

Semi-customized phantom for verification of conformal plans ● J. Y. GIRAUD et al.

185

Fig. 2. Prostate DVHs for the optimized treatment plan, the cross technique and the box technique.

der, and rectum were copied from the patient data to appropriate positions in the phantom. The target volume is the prostate and the seminal vesicles. Bladder and rectum are considered organs at risk. We carried out the optimization procedure to determine the optimized plan. This plan was evaluated along with the two conventional plans using the external geometry of the phantom and the internal anatomy of the patient. We compared the three treatment plans using dose–volume histograms (DVHs). For validation of the dosimetry of the optimized

plan, the phantom was filled with the Fricke-infused solution and irradiated according to the optimized plan. To generate a conversion curve, we filled five test tubes with the solution. These test tubes were irradiated with doses from 1 to 13 Gy. The phantom and the test tubes were read 20 minutes after irradiation with the MRI system. Magnetic resonance images are transferred to our system. The conversion curve was used to correlate gray levels of the MR images to relative dose. This allowed comparison between the

Fig. 3. Bladder DVHs for the optimized treatment plan, the cross technique and the box technique.

186

Medical Dosimetry

Volume 24, Number 3, 1999

Fig. 4. Rectum DVHs for the optimized treatment plan, the cross technique and the box technique.

measured dose distribution and the distribution calculated by the planning system. RESULTS The optimized plan consisted of seven asymmetric fields. They are described in Table 1. Figures 2, 3 and 4 show the DVHs for the prostate, the bladder, and the rectum resulting from the optimized plan, the box technique, and the cross technique, respectively. In Fig. 5, calculated isodoses are shown on a CT slice of the

phantom. The CT slice is on the beam axis. The calibration curve is plotted in Fig. 6. The linear interpolation is in a confidence interval of 1.4%. With this curve it is possible to obtain the T1 values that correspond to the calculated isodoses. Figure 5 shows the beam axis in white. Isodoses represented in this figure correspond to a percentage of the dose at the isocenter. The pink, blue, yellow, and green curves represent 38%, 51%, 78%, and 96%, respectively, of the dose at the isocenter. In this image we

Fig. 5. Result of the optimized treatment plan.

Semi-customized phantom for verification of conformal plans ● J. Y. GIRAUD et al.

187

Fig. 6. Standard curve: 1/T1 as a function of dose, obtained with an MRI system with Fricke-infused solution test tubes.

have also represented the outline of the prostate (red), the bladder (yellow), and the rectum (green). DISCUSSION In the optimized plan, the 96% isodose includes a part of the rectum. This is due to the inclusion of the seminal vesicles in the definition of the target volume, which results in an irregularly shaped target volume. Treatment of the seminal vesicles results in a significantly increased dose delivered to the rectum. It is notable that the optimization process removes the anteropos-

terior fields. These fields delivered a direct dose to the bladder and rectum. As a result, our optimization technique produces a plan for treatment of the prostate and seminal vesicles, which delivers less dose to these critical structures. On DVHs, it is noticeable that the box technique and the cross technique are similar even if the cross technique spares somewhat more bladder tissue. The optimized treatment plan is more inhomogeneous for the prostate than the others but results in greater sparing of the rectum and bladder than the other plans. The isogray levels in Fig. 7 correspond to the iso-

Fig. 7. Magnetic resonance image of the Fricke-infused solution and isogray level curves.

188

Medical Dosimetry

doses in Fig. 5. We observe that there is a good correlation between these curves. CONCLUSIONS Radiation therapy is becoming more and more conformal in delivery of dose to the target volume. The results of this study support continued development of optimization methods. The Fricke gel dosimeter is suitable for evaluation of three-dimensional distributions of dose delivered during real treatment. The comparison of an optimized plan with different conventional plans shows a significant advantage in terms of tissue sparing. Although this study examined prostate treatment only, the optimization could be applied to other treatment sites. It is clear that a better optimization result could be obtained by conformal shaping of each treatment field. This could be efficiently implemented using a multileaf collimator. Our future plans include support for the multileaf collimator in the optimization software. In this case, we plan to use our phantom to carry out quality control on treatment technique.

Volume 24, Number 3, 1999

REFERENCES 1. Menguy, Y.; Cinquin, P.; Laieb, N.; et al. Optimization in conformal therapy for prostatic cancer. Xith International Conference on Computers in Radiation Therapy, Manchester, UK March 20 –24, 1994. 2. Menguy, Y. Optimisation quadratique et ge´ome`trique de proble`mes de dosime´trie inverse. The`se de l’universite´ Joseph Fourier. Grenoble, France, 1996. 3. Rousseau, J.; Gibon, D.; Sarrazin, T.; Doukhan, N.; Marchandise, X. Magnetic resonance imaging of agarose gel phantom for assessment of three-dimensional dose distribution in linac radiosurgery. Brit. J. Radiol. 67:646 – 648; 1994. 4. Rousselle, I.; Rousseau, J.; Gibon, D.; Sarrazin, T.; Marchandise, X. MRI assessment of 3D dose distribution in radiosurgery using Fricke-infused, BANANA and PAAF gels. Society of Magnetic Resonance, Nice, France, 1995, p. 1056. 5. Olsson, L.E.; Petersson, S.; Ahlgren, L.; Mattsson, S. Ferrous sulphate gels for determination of absorbed dose distributions using MRI technique: basic studies. Phys. Med. Biol. 34:43–52; 1989. 6. Olsson, L.E.; Fransson, A.; Ericsson, A.; Mattsson, S. MR imaging of absorbed dose distributions for radiotherapy using ferrous sulphate gels. Phys. Med. Biol. 35:1623–1631; 1990. 7. Gore, J.C.; Kang, Y.S.; Schulz, R.J. Measurement of radiation dose distribution by nuclear magnetic resonance (NMR) imaging. Phys. Med. Biol. 29:1189 –1197; 1984.