An Artificial Intelligence Dose Engine for Fast Carbon Ion Treatment Planning
A. Quarz et al
The article presents the development of an artificial intelligence (AI)-based system designed to accelerate dose calculation in carbon ion radiotherapy. Carbon ion therapy is an advanced form of particle therapy that offers highly precise dose delivery and increased biological effectiveness against resistant tumors. However, treatment planning for carbon ions is computationally complex and time-consuming because it requires accurate modeling of particle interactions and tissue heterogeneity. The study explores whether AI can provide fast and reliable dose predictions while maintaining clinical accuracy.
The development of the research focused on training deep learning models using large datasets of previously calculated carbon ion treatment plans. The AI engine was designed to predict three-dimensional dose distributions directly from patient imaging and treatment parameters. Researchers compared AI-generated dose calculations with conventional Monte Carlo and analytical methods, which are considered highly accurate but require substantial computation time. Performance evaluation included dose distribution agreement, target coverage, organ-at-risk sparing, and calculation speed.
The results showed that the AI dose engine achieved excellent agreement with standard clinical dose calculation methods. Dose differences were minimal, and target coverage remained clinically acceptable across a wide range of treatment scenarios. Importantly, the AI system reduced calculation times dramatically, enabling near real-time dose estimation. This improvement could significantly streamline adaptive treatment planning workflows and facilitate more efficient clinical implementation of carbon ion therapy.
In conclusion, the article demonstrates that artificial intelligence has strong potential to transform carbon ion treatment planning by combining speed with high dosimetric accuracy. The proposed AI dose engine may improve workflow efficiency, support adaptive radiotherapy strategies, and increase accessibility to advanced particle therapy techniques. The authors emphasize that further clinical validation and multicenter testing are necessary before widespread routine implementation can be achieved.
Publisehd by the International Journal of Particle Therapy 19 (2026) 101309
