Session Item

To implement KB-based automatic planning for Helical-Tomotherapy (Accuray Inc., HTT) by using a commercial software available out of the HTT planning station (TPS). Focus of the first clinical implementation was the case of high-risk prostate cancer, including pelvic nodes irradiation in a SIB approach.

Our clinical protocol consists in delivering 74.2Gy to prostate and proximal seminal vesicles (PTV_high), 65.6Gy to cranial portion of seminal vesicles (PTV_int) and 51.8Gy to pelvic lymph nodes (PTV_LN) in 28 fractions. One-hundred-two HTT clinical plans were selected to train a KB-model using the RapidPlan (RP) tool implemented in the Eclipse TPS (v 13.6, Varian Inc.). RP is configured to model plans delivered with VMAT - RapidArc (RA) plans. Hence, all plans were exported from the HTT-TPS to Eclipse and linked to virtual RA-plans. The resulting KB-model was interactively fine-tuned in terms of statistical DVH-estimation and optimized template for the optimization, aiming at maximizing its robustness. Then, an internal (20 patients inside the model) and an external validation (30 new patients) were performed to assess the performances of the model. All automatic HTT-plans (KB-TP) were compared against the original plans (TP) in terms of OARs/PTVs dose-volume parameters. Wilcoxon-tests were performed to assess statistically significant differences (p < 0.05). To automatize the entire HTT-planning workflow, the individually optimized KB-based templates are converted in HTT-like template and sent automatically to the HTT-TPS through scripting. The individual template is then associated to the patient in the HTT-TPS and the full dose calculation is set after 300 iterations, without any additional planner intervention (Figure 1).

KB-TP plans were generally better than or equivalent to TP plans, in both validation cohorts (Figure 2). PTVs coverage were comparable for the internal sets, meanwhile PTV_high and PTV_int coverages were slightly improved for the external ones. Moreover, a significant improvement in PTVs and OVERLAP (between rectum and PTV_high) homogeneity were observed for both set. OARs sparing for KB-TP was slightly improved, more evidently in the external validation group. Of note, V_20Gy, V_40Gy and D_max for the bladder were significantly better in KB-TP plans, V_20Gy and D_mean for the bowel, as well as for V_68Gy and D_max of the rectum. The automatic KB-based technique was successfully implemented in the clinical routine with a consequent large reduction of the planning time and better plan homogeneity, hopefully avoiding any risk of sub-optimal plans.

We demonstrated the feasibility of the clinical implementation of the KB-approach to treatments delivered with HTT. For the first time to our knowledge, fully automatic planning workflow was successfully implemented for HTT planning optimization. The KB-based planning approach was able to generate high-quality automatic HTT plans, without any intervention of the planner.