Knowledge-based versus deep learning based treatment planning for breast radiotherapy

Abstract

Background and PurposeTo improve radiotherapy (RT) planning efficiency and plan quality, knowledge-based planning (KBP) and deep learning (DL) solutions have been developed. We aimed to make a direct comparison of these models for breast cancer planning using the same training, validation, and testing sets. Materials and MethodsTwo KBP models were trained and validated with 90 RT plans for left-sided breast cancer with 15 fractions of 2.6 Gy. The versions either used the full dataset (non-clean model) or a cleaned dataset (clean model), thus eliminating geometric and dosimetric outliers. Results were compared with a DL U-net model (previously trained and validated with the same 90 RT plans) and manually produced RT plans, for the same independent dataset of 15 patients. Clinically relevant dose volume histogram parameters were evaluated according to established consensus criteria. ResultsBoth KBP models underestimated the mean heart and lung dose equally 0.4 Gy (0.3–1.1 Gy) and 1.4 Gy (1.1–2.8 Gy) compared to the clinical plans 0.8 Gy (0.5–1.8 Gy) and 1.7 Gy (1.3–3.2 Gy) while in the final calculations the mean lung dose was higher 1.9–2.0 Gy (1.5–3.5 Gy) for both KPB models. The U-Net model resulted in a mean planning target volume dose of 40.7 Gy (40.4–41.3 Gy), slightly higher than the clinical plans 40.5 Gy (40.1–41.0 Gy). ConclusionsOnly small differences were observed between the estimated and final dose calculation and the clinical results for both KPB models and the DL model. With a good set of breast plans, the data cleaning module is not needed and both KPB and DL models lead to clinically acceptable results.