An algorithm has been developed to calculate the collimator jaw motions required to generate intensity-modulated fields far use in conformal radiotherapy. The dynamic technique allows arbitrary intensity profiles to he generated using a single unidirectional sweep of the collimators. The collimator jaws have independent motion, so that an aperture of variable width is scanned across the field. The algorithm has the form of a constrained optimization problem and jaw motions are optimized such that the treatment time for the field at a given dose rate is minimized. The application and results of this technique are presented, and it is shown that this approach provides an efficient practical implementation of conformal radiotherapy plans based on the use of intensity-modulated fields. The technique can be extended to 3 D treatment plans and fields through the use of computercontrolled multileaf collimators.

The importance of geometric and dosimetric inaccuracies in the delivery of conformal radiotherapy using intensity-modulated x-ray fields is assessed. It is shown that although the incident intensity or fluence profiles may have large gradients, the resulting absorbed dose profile gradients are shallower because of electron transport and photon scatter. This makes the interleaving of dose contributions from the many intensity-modulated fields used less critically dependent on their relative positions and on the accuracy of profile generation. It is found that the expected geometric and dosimetric inaccuracies do not make this technique impracticable provided that good patient fixation and beam delivery can be assured.

A small computer has been interfaced to a linear accelerator used for routine radiotherapy. The objectives are to assess the rate and magnitude of some of the mistakes which may occur, to prevent these where possible and to assist radiographers with record keeping. In this paper we describe our approach to the problem, the apparatus used, the computer program, and the results of several years of operating experience while the system was continually being developed. On one linear accelerator, we estimate that an error has been prevented roughly once in every 500 field parameters set. The most common error is in setting the diaphragms and a detailed analysis is given of possible reasons. Doses in monitor units are recorded automatically and a running total of cumulative dose is maintained; plans for the printing of the dose in rads at selected points are discussed.

No abstract

No abstract