Click here to

Session: Treatment Planning: Planning Automation and Assessment [Return to Session]

Knowledge-Based Cardiac Sparing Treatment Planning for Lung Radiotherapy

J Harms*, J Zhang, O Kayode, J Wolf, S Tian, N McCall, K Higgins, R Castillo, X Yang, Winship Cancer Institute of Emory University, Atlanta, GA

Presentations

SU-F-TRACK 5-5 (Sunday, 7/25/2021) 4:30 PM - 5:30 PM [Eastern Time (GMT-4)]

Purpose: High radiation doses to the heart are correlated with poor overall survival in radiotherapy patients with stage III non-small cell lung cancer (NSCLC). Cardiac sparing is an optimization goal of all lung treatment plans, but optimization based on the whole heart has limitations. Optimization of treatment plans using cardiac substructures can reduce cardiac dose, but is not feasible in most clinics due to the extra time needed for planning. This study builds a knowledge-based cardiac sparing planning (KBCSP) model for cardiac dose reduction using a set of treatment plans for patients with stage III NSCLC.

Methods: Cardiac substructures were delineated for 33 patients and treatment plans were re-optimized to reduce dose to substructures and the heart. A KBCSP model was developed from these plans using target volumes and organs-at-risk (OARs) including the heart but excluding substructures. To test the model’s efficacy for cardiac dose reduction, a second KBP model was trained from the same 33 patients, using the original clinical treatment plans. Both models were evaluated in a cohort of 30 separate stage III NSCLC patients.

Results: Both models produced acceptable target coverage, dose uniformity, and dose to OARs. Compared to the KBP-based plans, KBCSP-based plans showed significant reductions in mean dose to the esophagus and lungs while performing similar or better in all evaluated heart dose metrics. Compared to the clinical KBP model, the KBCSP model showed reduced (p<0.05) heart mean and maximum doses, as well as volumes receiving 60 Gy, 50 Gy, and 30 Gy.

Conclusion: The KBCSP model successfully captured the cardiac-sparing expertise and improved plan quality for patients in a validation cohort. This KBCSP model offers an automated approach to reduced cardiac dose without explicit delineation of cardiac substructures, which may reduce heart toxicity and improve overall survival in radiotherapy patients with NSCLC.

Handouts

    Keywords

    Not Applicable / None Entered.

    Taxonomy

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

    Contact Email

    Share: