Using Design of Experiments to Optimise Planning Parameters for SBRT Lung Single Iso Multi-Lesion

Purpose or Objective: This study utilises the design of experiments to evaluate the VMAT planning parameters of single isocenter treatment plans for multiple lung metastases. An optimization model based on Taguchi and Principal component analysis is employed to optimise the planning parameters including: arc arrangement, calculation grid size, calculation model, and beam energy on multiple performance characteristics, conformity index and lung V5Gy.

Material and Methods: Treatment plans, each with 2 metastatic lung lesions, were planned using single isocenter technique. The collimator angles were optimised to avoid open areas. In this analysis, four planning parameters (a-d) were considered: (a) Arc arrangements: set1: Gantry 181cw0, coll 330; gantry0ccw181, coll 30. set2: Gantry 0cw179, couch15 Gantry179cw0, couch15; (b) Energy: 6-MV-FFF, 10MV-FFF; (c) Calculation grid size: 1mm, 1.5mm (d) Calculation models: AAA, Acuros. Treatment planning was performed in Varian Eclipse (ver.13.7). A suitable orthogonal array was selected (L8) to perform the experiments. After conducting the experiments with the combinations of planning parameters, the conformity index (CI) and lung V5Gy S/N ratio for each parameter was calculated. Optimum levels for the multiple response optimizations were determined.

Results: We determined that the factors most affecting the conformity index are arc arrangement and beam energy. These tests were also used to evaluate lung V5Gy. In these evaluations, the significant parameters were grid size and calculation model. Using the utility concept, we determined the combination of each of the four parameters tested in this study that yielded the best quality plans: (a) arc arrangement-set2, (b) 6MV-FFF, (c) calc.grid 1mm, (d) Acuros algorithm. Of these, the dominant significant influences on plan quality are arc arrangement, and beam energy.

Conclusion: Results were analysed using ANOVA and were found to be within the confidence interval. Further investigation using this methodology are recommended. Such parameters might include: Dose to whole lung, esophagus, heart, spinal cord, bronchial tree, and pulmonary artery.