Click here to

Session: Therapy: External Beam: Automatic Treatment Planning [Return to Session]

A Fully-Automated Field-In-Field Algorithm Translatable to Multiple Disease Sites

K Huang1,2*, D Rhee1,2, A Olanrewaju2, D Hancock2, L Zhang2, C Cardenas1,2, P Das2, S Beddar1,2, D Fuentes1,2, T Briere1,2, L Court1,2, (1) University of Texas MD Anderson UTHealth Graduate School of Biomedical Sciences, (2) University of Texas MD Anderson Cancer Center, Houston, TX


MO-IePD-TRACK 5-6 (Monday, 7/26/2021) 12:30 PM - 1:00 PM [Eastern Time (GMT-4)]

Purpose: To develop an adaptive automated field-in-field technique compatible with various beam arrangements and treatment sites.

Methods: We created an algorithm that automates the clinical workflow for creating field-in-field plans independent of treatment planning system. The algorithm automatically identifies a hotspot, creates a subfield, calculates dose, and optimizes beam weight without user intervention. This process is repeated until the hotspot is reduced. Parameters used in the algorithm are configurable and include the definition of hotspot, the target volume, the maximum number of subfields, the minimum MU per field, and the optimization solver, and so on. A sub-field is determined based on hotspot area. The beam weights are optimized based on user-configured physical constraints for DVH coverage and least-squared cost functions. The final plans were evaluated based on target volume coverage and the remaining hotspot dose. The algorithm was tested on 20 patients each for whole brain (2-field), rectal (3-field), and cervix (4-field-box) with the clinically delivered primary fields as inputs.

Results: For whole brain plans, all plans achieved V98.5% with hotspot less than 108% of prescription dose. For rectal plans, all plans achieved V95% with hotspot less than 110% of prescription dose. For cervix plans, all plans achieved V97% with hotspot less than 109% of prescription dose. On average, the algorithm reduced the hotspots of the plans from 114% to 107%, 130% to 107%, and 111% to 107% for whole brain, rectal, and cervix, respectively. The average time to complete a plan is 8.5 mins, 22.2 mins, and 6.5 mins for whole brain, rectal and cervix plans, respectively.

Conclusion: We automated the clinical workflow for generating FIF to reduce hotspots in 3D conformal plans. This algorithm successfully produced FIF plans for beam arrangements and multiple disease sites and can improve treatment planning efficiency when incorporated into automatic planning workflow.



    Radiation Therapy, Treatment Planning, Treatment Techniques


    TH- External Beam- Photons: Treatment planning using machine learning/Knowledge Based Planning/automation

    Contact Email