ESTRO 2024 - Abstract Book

S4428

Physics - Machine learning models and clinical applications

ESTRO 2024

the right coronary (+8Gy), and underestimated the maximum dose to the left breast (-4.5Gy). The post-processing was used to reduce the predicted dose for these OARs (except for the left breast) below their corresponding values of the manual plans accounting for the typical dose deterioration occurring after mimicking. As expected, the mimicking process deteriorated the post-processed predicted dose while maintaining plan quality similar to the manual cases. During model validation, the median values of the 15 plans fulfilled the dose-volume requirements for both manual and auto-plans. A few significant differences, yet not clinically relevant, were observed between manual and auto-plans, particularly in the OARs sparing. The right lung and the right coronary were more spared in auto plans. In contrast, the maximum dose for the left lung was lower in manual plans (see Table 1).

Conclusion:

When applied to right-sided patients, the left-sided auto-planning model’s dose predictions required substantial corrections before mimicking. The post-processing of the predicted dose effectively adapted the dose for the right breast planning requirements. The auto-planning process for right-sided breast treatments was successfully implemented, yielding dose distributions of comparable clinical quality.