Real-time motion modeling and treatment verification for irregular motion in carbon ion therapy: a feasibility study

Objective. Irregular motion impacts treatment accuracy and can be compensated by larger margins or online adaptive approaches. A seamless workflow for fast and accurate 4D-dose reconstruction allows dosimetric monitoring intra- and inter-fractionally, as a basis for adaptive therapy. This study presents a real-time, motion-adaptive framework that combines motion modeling and treatment verification, integrated into the dose delivery and monitoring systems to enable continuous assessment of the delivered 4D-dose. Approach. The framework includes a GPU-based analytical algorithm for real-time dose reconstruction in carbon ion therapy, interfaced with the dose delivery and optical tracking systems at the Centro Nazionale di Adroterapia Oncologica (CNAO). A motion model, driven by external surrogate tracking, generates a virtual CT every 150 ms, used for 4D-dose reconstruction with measured spot parameters. Planned and delivered doses are compared after each iso-energy slice. The framework was validated at CNAO using a geometric target and a 4D lung tumor phantom with a moving 2D ionization chamber array, under regular and irregular motion patterns. Main results. The framework successfully generated real-time CT images of the lung phantom, showing strong agreement with ground-truth images. Dose reconstructions were performed within inter-spill times during delivery, ensuring rapid assessment. Comparisons against detector measurements yielded an average gamma-index passing rate of 99% (3%/3 mm), confirming the accuracy of both the motion model and the integrated treatment verification system. Significance. This work presents the first real-time framework for carbon ion therapy, integrating motion modeling and dose reconstruction to handle irregular motion, fully embedded in a clinic-like setup.