Physician-driven artificial intelligence enabled planning for intraprostatic dose escalation in under ten minutes.

187 Background: Intraprostatic radiation dose escalation is an area of clinical interest. Dose escalation within the prostate must be balanced with maintaining acceptable dose to the organs at risk, OAR (bladder, rectum, and urethra). Treatment planning therefore requires simultaneous consideration of multiple competing plan optimization goals, for which iterative, interdisciplinary treatment planning tasks may take significant physician, physicist, and dosimetrist time. Semi-automated treatment planning using artificial intelligence has the potential to significantly reduce treatment planning time for technically complex treatments. Methods: A prostate SBRT planning template was created using the Varian ETHOS treatment planning system (TPS) combined with an in-house RapidPlan SBRT prostate model. Prostate dose was prescribed to 36.25 Gy over 5 fractions with 95% coverage to the PTV. To respect standard SBRT normal tissue toxicity constraints while simultaneously escalating intraprostatic dose, the TPS automatically created an intraprostatic boost structure (PTV_SIB), derived from the PTV by excluding OARs with a pre-determined margin. Physicians were trained to perform treatment planning using the prostate SBRT planning template. Treatment planning was performed on 5 unique patients. The time spent from initiation to end of treatment planning and dosimetric parameters were recorded. Results: For each patient, the ETHOS TPS generated two SBRT plans (9 field static IMRT and 3 VMAT arc) with intraprostatic dose escalation in an average of 9.3 minutes [range 8.4-11.8]. Static field and VMAT plans were comparable. PTV_SIB was escalated to above 50 Gy in all cases. Relevant dosimetry for each patient’s static IMRT plan is shown. Conclusions: Physician-driven ETHOS treatment planning was able to produce boosted internal PTV doses using autosegmented volumes. The ETHOS TPS was able to generate dose-escalated plans that reconciled complex OAR and PTV goals within 8-12 minutes. Hence, the ETHOS TPS opens the possibility of rapid physician-driven treatment planning throughput. [Table: see text]