Hui Yan, Presenter: Nothing to Disclose

Purpose/Objective: Beam angle selection is one of important parameters in inverse treatment planning. Currently, orientation for each IMRT beam is set based on the experience in almost all commercial planning systems, which may potentially degrade the quality of IMRT treatment, especially with smaller number of beams. To automatically and optimally determine the beam angles for an inverse plan, an artificial intelligent approach for the beam angle optimization is being investigated. Our expectation was that with the help of a fuzzy inference system, an optimal combination of beam angles for each inverse plan could be automatically selected with a built-in heuristic searching strategy. Materials/Methods: Prior to the beam angle optimization, a look-up table counting the relative coverage area of the target, the critical organs, and the normal tissue on the radiation path of each beam angle (0-360o) were created. Based on it, three other tables (Table_TV, Table_CO, Table_NT) were generated by re-sorting the content of look-up table by the coverage volumes of target, critical organ, and normal tissue, respectively. Each time, while the dose distribution was obtained using the pre-set beam angles, the dose volume histograms (DVH) of interested organs within the coverage volume of each beam were achieved and compared to their corresponding DVHs prescribed. The differences between them were calculated and acted as the input parameters to the fuzzy inference system (FIS), which will suggest modifications of coverage volume for each interested organs based on the differences between calculated and prescribed DVHs. After the outputs of FIS were achieved, the next candidate beam angles were selected from the Table_TV, the Table_CO, and the Table_NT. And the final beam angle was determined from maximal agreement between these candidate beam angles. Results: To evaluate the beam angle optimization method using fuzzy inference technique, several simulated cases were tested using this technique and their results (DVHs) were compared to the outputs from plans generated with equal beam angle setups. For most of cases, the dose distributions achieved by plans with optimized beam angles exceeded the performance of plans without beam angle optimization or equal beam angle setup. On average, 10-20% reduction of critical organ dose and more than 10% reduction of normal tissue dose were achieved, while there was no apparent degradation of target dose. Conclusions: The fuzzy inference technique could be used for beam angle optimization in the treatment inverse planning. The improvement of dose distribution is apparent, and the outcome is always better or equal than equal-beam angle setup. The technique is potentially useful to determine the optimal beam number.

Yan, H, Beam Angle Optimization Using Fuzzy Inference System in an Inverse Planning System.  Radiological Society of North America 2004 Scientific Assembly and Annual Meeting, November 28 - December 3, 2004 ,Chicago IL. http://archive.rsna.org/2004/4417880.html