Proceedings of the 39th r\nnual ASTRO Meeting 183

96

TREATMENT IMPROVEMENT POTENTIAL WITH HIGHER DOSE ACCURACY FOR LUNG TREATMENTS

R. Mohan,l,Z Q. Wu,lv2 C. Hartmann-Siantar,3 W. Chandler Ullrichl,2

,3 P. Garmonb2, 0. Tercilla, T. Huang and R. Schmidt-

lMCV Hospitals and 2McCuire VA Hospital, Richmond, VA; 3Lawrence Livermore National Lab., Livermore, CA

Puruosa: The purpose of this investigation is to determine if higher accuracy of predicted dose distributions for treatments involving lungs and other low density inhomogeneities, attained through Monte Carlo simulations of radiation transport, will reveal clinically significant differences between the “true” dose distribution and the one predicted by the empirical dose computation models implemented on commercial and non-commercial treatment planning systems in clinical use. Mainly because of the inability of the conventional dose computation models to adequately account for the lateral transport of radiation, there may be substantial discrepancy between the computed dose and actual dose. Measurements in phantoms have shown significant discrepancies, the magnitude of which depends upon a number of factors including the density and size of the inhomogeneity, distance from the field boundary and distance from the interface. However, because of the lack of tools to predict dose accurately, the clinical consequences of dose discrepancies have only been a matter of speculation.

-and Monte Carlo simulation of radiation transport is the most accurate, essentially exact, means of predicting dose. However, because of the complexities in implementing this technique into clinical practice and the enormous amount of computer time required, it has hitherto fore been considered to be impractical. Computer speed has improved dramatically over the years. More importantly, through clever algorithm improvements and variance reduction techniques, the Monte Carlo methods have already been shown to be within the realistic range and implementation for routine clinical use is within reach. The most advanced such methods is PEREGRINE, a Monte Carlo dose computation engine being developed at Lawrence Livermore National Lab. Accuracy of PEREGRINE is being confirmed in separate studies. We used PEREGRINE to compute 3D conformal treatment plans for lung patients and compared them with corresponding treatment plans obtained using a conventional dose computation model. In the near future, we also plan to compare results with the non-local energy deposition (NLED) model implemented on the Pinnacle treatment planning system. This type of model is partially able to take lateral transport into account. Plan evaluation and comparison was based on dose indices such as minimum and maximum tumor doses, dose-volume histograms, the volume of tumor receiving less than the prescription dose and the volume of normal tissue within the prescription isodose surface. This information was supplemented with relative values of tumor control probabilities and normal tissue complication probabilities to facilitate comparison of inhomogeneous dose distributions.

&&: Dose distributions computed with Monte Carlo simulations were found to be considerably different and more inhomogeneous as compared to the corresponding dose distributions based on conventional methods. For example, for a 5 beam, 6 MV 3D conformal lung plan, while the 70 Gy prescription line appeared to cover 95% of the planning target volume in conventional dose distribution, in reality only the 60 Gy isodose line did. Small fractions of the planning target volume were found to have as much as 30% under-dosing caused by the lost of electronic equilibrium in regions near boundaries of beams traversing through low density media. Minimum and maximum target doses for the conventional plan were 68 and 81 Gy respectively, but Monte Carlo calculations showed that they were. 50 Gy and 70 Gy instead. These small low dose regions, although not discernible on tumor DVHs, may have significant effect on the tumor control probability. For the example mentioned, the TCP computed using Goitien’s model was 75% for the conventional dose distribution as compared to 64% for PEREGRINE.

Conclusions: Our investigation is providing increasing evidence that accurate computations produced with Monte Carlo simulations will reveal clinically significant discrepancies of treatment plans based on conventional dose computation models. A systematic study of a number of cases is needed to establish the advantages of Monte Carlo methodology. conventional treatment plans.

Furthermore, strategies need to be developed to rectify the weaknesses discovered in

97

A COMPUTER-CONTROLLED HIGH RESOLUTION MICRO-MULTI-LEAF COLLIMATOR FOR STEREOTACTIC CONFORMAL RADIO- THERAPY

Wolfgang Schlegel, Otto Pastyr, Rudolf Kubesch, Torsten Diemer, Gunnilla Kiister, Bernhard Rhein and Karl-Heinz Htiver

Deutsches Krebsforschungszentrm (dkfi), Department ofMedical Physics, Im Neuenheimer Feld 280, D-69121 Heidelberg, Germany

Purpose/Objective In stereotactic conformal radiotherapy of irregularly shaped lesions, either multi-isocentric convergent beam treatment techniques with circular collimators or irregular shaped beams are being used While the treatment technique with multiple isocenters has the disadvantage of producing inhomogeneous dose distributions, the use of irregular shaped fields is not yet satisfying from a technical point of view: Cerrobend blocking or the use of static micro MLCs need a long preparation time and only allow static treatment techniques, MLC collimators which are commercially available in connection with modem LINACs have leaf-thickness of at least 1 cm which is too coarse for stereotactic radiotherapy of lesions in the brain and head and neck area For this reason, we developed a computer controlled micro&EC with technical specifications matched to the needs of stereotactic radiotherapy and radiosurgery.

Materials and Methods The mechanical specifications of the computer controlled micro-MLC were derived from our experience with stereotactic treatment techniques, from the requirement that the MLC has to be attachable as an external device to the accessory holders of standard LlNACs, simetric measurements as well as Monte Carlo calculations.

including cost considerations, do-

The Micro-MLC is controlled by an electronic equipment consisting of a standard PC under Windows 95, an interface board, 14 Micro-controller boards, a verification system and 80 driving units equipped with DC motors and potentiometers. The control program has calibrating, operating, visualizing and test options Irregular field data are transferred from the treatment planning computer to the control PC and distributed to the micro-controllers, which in parallalel are driving three leaves each. Beside the special control unit, we are currently investigating whether the electronics of commercially available integrated large field MLCs can also be used for operating the Micro-MLC.

Results The micro-WC consists of 40 tungsten leaf pairs of Imm thickness, thus reaching a leaf width of I 6 mm and a maximum field size of 73 x 64 mm at isocenter distance There is a maximum overtravel of 24 mm and a max speed of 1.5 cm/set. The mechanical and electronical tests which we performed with the prototype Micro-MLC showed, that the time needed to move all leaves from fblly open to fitlly close is approximately 2 seconds Repeatability and absolute positioning data arc in the range of 0.2 mm Penumbra, interleaf leakage penetration were investigated using film dosimetry and Monte Carlo calculations The results (penetration 0.5 % at 15 MV X- rays, mean interleaf leakage 0.8%. max interleaf leakage 2.5% ) arc in good agreement with the design specifications,

Conclusion With the motor driven Micro-MLC we have designed and developed a beam shaping device which will significantly facilitate and improve stereotactic treatment techniques for lesions in the brain and head and neck area The system meets all requirements of conformal treatment techniques, including mul- tiple static irregular shaped fields, dynamic field shaping as well as intensity modulated beams Due to the flexibility and rhe modularity of the design, the system offers an optimal basis for the development of accessory-MLCs with larger field sizes.