Program Information
Genetic Algorithm Based Script for Planning Automation: Preliminary Results for Prostate Cancer
E Gallio1*, A Alparone2 , C Fiandra3 , C Vecchi4 , G Balestra5 , R Ragona6 , U Ricardi7 , (1) A.O.U. Citta' della Salute e della Scienza, Turin, TO, (2) Politecnico di Torino, Torino, TO, (3) University of Torino, Torino, TO, (4) Tecnologie Avanzate, Torino, TO, (5) Politecnico di Torino, Torino, TO, (6) Univ Turin, Turin, ,(7) University of Torino, Torino, TO
Presentations
SU-E-FS2-5 (Sunday, July 30, 2017) 1:00 PM - 1:55 PM Room: Four Seasons 2
Purpose: To automate optimization process in RayStation treatment planning system (TPS).
Methods: A Python script based on genetic algorithm was implemented for VMAT treatment planning of prostate tumor. The chromosomes of the algorithm were max equivalent uniform dose functions (maxEUDs) for rectum and bladder. Three initial couples were randomly generated and a maximum of five reproductions were permitted. The fitness function included the root mean square of conformity index (CI) of planning target volume (PTV) and mean dose of bladder, rectum and femoral heads with different weights. The script was tested for two different clinical prescriptions: five patients with 78 Gy to PTV and five patients with simultaneous integrated boost (70.2 Gy and 61.1 Gy). A comparison with corresponding plans created with Monaco TPS (M, Elekta) and Auto-Planning module of Pinnacle3 (AP, Philips) was carried out. The plans were evaluated with a score of PlanIQ (Sun Nuclear) in terms of CI and constraints of organ at risks.
Results: All the plans respected the target coverage (PTV: V95% = 95%, V107% < 2%). Table 1 reports the scores of the first (a) and second (b) group of patients. Script plans scores were always higher to Monaco ones with average total scores differences (ASD) of 20.6±15.9 and 11.1±7.8 for first and second group respectively. In comparison to Auto-Planning plans, for the first group of patients, script total scores were higher for three times out of five (ASD = 8.1±20.7); whereas for the second patients groups, Auto-Planning total scores were always better (ASD = -9.5±9.0).
Conclusion: Automatic planning gives better results than manual ones. The plans generated by the implemented genetic algorithm based script were clinically comparable with commercial solution AP. Future developments will be aimed to cover other clinical districts.
Contact Email: