A Novel Image-Based Framework for Real-Time 3D Motion Tracking of Abdominal Tumors on Conventional Linear Accelerators
Abstract
Purpose
Radiotherapy for upper abdominal cancers is limited by respiratory motion and the low radiation tolerance, restricting safe dose escalation. Conventional linear accelerators rely on kV X-ray and CBCT imaging but lack real-time internal motion tracking capability, preventing reliable delivery of ablative treatments. This work aims to develop a non-invasive, image-based framework for continuous real-time 3D abdominal motion tracking on conventional Linac systems.
Methods
Our abdominal motion tracking workflow consisted of both pre- and on- treatment-day components. Before treatment, compact and patient-specific 4D respiratory motion models were derived from 4DCT using deformable image registration (DIR) and principal component analysis (PCA). To generate realistic free-breathing motion beyond 4DCT, additional patients’ respiratory signals were applied to drive the PCA model to synthesize continuous motion-deformed CT volumes and time-resolved DRRs, paired with ground-truth motion and gantry rotation coefficients. The motion prediction models were pre-trained on these synthesized DRRs to predict PCA motion coefficients. On the treatment day, the pre-trained models were refined using AI-enhanced 4D-CBCT and X-ray fluoroscopy images. The PCA coefficients predicted using real-time X-ray fluoroscopy images acquired during treatment delivery were used to efficiently reconstruct continuous 3D internal tumor motion for treatment guidance.
Results
The 4DCT motion models accurately represented respiratory deformation, with three PCA components approximating dense abdominal motion at 0.1 ± 0.2 mm reconstruction error. The proposed prediction framework reliably recovered continuous 3D motion across diverse breathing scenarios, achieving less than 0.5% relative error on PCA coefficient prediction and a mean motion prediction error of 0.19 ± 0.11 mm.
Conclusion
This work presents a comprehensive, clinically compatible image-based workflow for real-time 3D abdominal motion tracking on conventional linear accelerators. The demonstrated accuracy and efficiency could support meaningful margin reduction, reduced gastrointestinal toxicity, and safe delivery of ablative doses, with the potential to substantially improve abdominal radiotherapy outcomes. .