OPTIMIZATION OF 3D CONFORMAL TREATMENTS

点击翻译按钮获取中文摘要

It is becoming evident that three-dimensional radiation treatment planning (3DRTP) is essential for achieving higher cure rates. Early experience with 3DRTP has demonstrated its potential as well as limitations. A major difficulty in 3DRTP is that the planning data are voluminous, displays are difficult to comprehend and the number of options to be explored are greatly increased. Consequently, design and evaluation of plans is difficult and it is virtually impossible to explore a sufficiently large domain of solutions by trial and error. Thus, full clinical potential of three-dimensional treatment planning cannot be realized without the application of computer-aided optimization. In order for the optimization process to be meaningful, it must incorporate clinical consequences of changes in dose distributions resulting from changes in beam characteristics. We hypothesize that, even though the models for predicting clinical consequences are simplistic and the clinical and biological data for making such predications are known to be sparse and not entirely reliable, incorporation of clinical and biological data into the optimization process will steer the solution of the treatment planning problem in the direction of lower normal tissue complications and higher tumor control. We propose to design and develop formalisms, algorithms and software for clinically relevant optimization of 3D conformal treatment plans. We also propose to develop new biological models. We will use dose distribution data, biological models and observed normal tissue and tumor response data to compute tumor control and normal tissue complication probabilities for each of the critical normal structures encountered in a treatment plan. These quantities will be combined into a single score using an objective function which will incorporate the importance of each end point as assessed by the physician. Using the "simulated annealing" method of optimization, the beam (or ray) weights will be adjusted to maximize the score. In general, additional constraints will need to be applied to ensure consistency of the results of optimization with the judgement of the physician. We propose to apply the optimization process to two classes of conformal treatment planning problems: (1) treatments consisting of one or more automatically delivered fields, each consisting of a number of multi-leaf collimator shaped conformal segments in which the optimum set of weights of fields and segments are determined, and (2) treatments consisting of a small number of intensity-modulated fields in which optimum fluence patterns are designed by adjusting "ray weights". These approaches have shown considerable promise in our preliminary investigations.