ESTRO 2024 - Abstract Book

S4435

Physics - Machine learning models and clinical applications

ESTRO 2024

[4] Stanley J, Breitman K, Dunscombe P, Spencer DP, Lau H. Evaluation of stereotactic radiosurgery conformity indices for 170 target volumes in patients with brain metastases. J Appl Clin Med Phys. 2011 Feb 1;12(2):3449. doi: 10.1120/jacmp.v12i2.3449. PMID: 21587193; PMCID: PMC5718686.

[5] Paddick I, Lippitz B. A simple dose gradient measurement tool to complement the conformity index. J Neurosurg. 2006 Dec;105 Suppl:194-201. doi: 10.3171/sup.2006.105.7.194. PMID: 18503356.

[6] Reynolds TA, Jensen AR, Bellairs EE, Ozer M. Dose Gradient Index for Stereotactic Radiosurgery/Radiation Therapy. Int J Radiat Oncol Biol Phys. 2020 Mar 1;106(3):604-611. doi: 10.1016/j.ijrobp.2019.11.408. PMID: 32014151.

[7] Grimm J, Marks LB, Jackson A, Kavanagh BD, Xue J, Yorke E. High Dose per Fraction, Hypofractionated Treatment Effects in the Clinic (HyTEC): An Overview. Int J Radiat Oncol Biol Phys. 2021 May 1;110(1):1-10. doi: 10.1016/j.ijrobp.2020.10.039. PMID: 33864823; PMCID: PMC9447432. [8] Videtic GM, Hu C, Singh AK, Chang JY, Parker W, Olivier KR, Schild SE, Komaki R, Urbanic JJ, Timmerman RD, Choy H. A Randomized Phase 2 Study Comparing 2 Stereotactic Body Radiation Therapy Schedules for Medically Inoperable Patients With Stage I Peripheral Non-Small Cell Lung Cancer: NRG Oncology RTOG 0915 (NCCTG N0927). Int J Radiat Oncol Biol Phys. 2015 Nov 15;93(4):757-64. doi: 10.1016/j.ijrobp.2015.07.2260. Epub 2015 Jul 17. Erratum in: Int J Radiat Oncol Biol Phys. 2016 Mar 1;94(3):638. Timmerman, Robert D [added]. PMID: 26530743; PMCID: PMC4744654. [9] Finitsis S, Anxionnat R, Bracard S, Lebedinsky A, Marchal C, Picard L. Symptomatic Radionecrosis after AVM Stereotactic Radiosurgery. Study of 16 Consecutive Patients. Interv Neuroradiol. 2005 Mar 17;11(1):25-33. doi: 10.1177/159101990501100104. Epub 2005 Jun 17. PMID: 20584432; PMCID: PMC3403785.

749

Digital Poster

Reinforcement learning for accelerated automatic treatment planning optimization

Eva Anjo 1 , Humberto Rocha 2 , Joana Dias 3

1 University of Coimbra, FCTUC, Coimbra, Portugal. 2 University of Coimbra, CeBER, Coimbra, Portugal. 3 University of Coimbra, INESC Coimbra, Coimbra, Portugal

Purpose/Objective:

Machine learning for automatic treatment planning requires training based on large datasets of already calculated high quality plans. These datasets are not always available and very good results can be hard to achieve especially if the particular case at hand deviates from the cases that constitute this dataset. In this work a combined approach merging Reinforcement Learning (RL) and Optimization is developed to overcome these disadvantages.

Material/Methods:

This work considered the automatic generation of IMRT treatment plans for prostate cancer cases, with the tumor localized within the prostate gland, post-surgery. The structures of interest included the rectum, bladder, left and