Motion correction in CBCT imaging using gate-less model-based reconstruction of non-rigid motion and images
Ethan Waterink1, Rodrigo José Santo1, Cornelis A T van den Berg1,2
1Department of Radiotherapy, Utrecht University and University Medical Center Utrecht, Utrecht, The Netherlands.
Medical Physics
|October 9, 2025
Summary
This study introduces CBCT-MOTUS, a novel gate-less method to correct patient motion during radiotherapy imaging. It effectively reduces motion artifacts, improving image quality for better treatment accuracy.
Area of Science:
- Medical Imaging
- Radiotherapy Physics
- Computational Imaging
Background:
- Cone beam computed tomography (CBCT) is vital for radiotherapy patient positioning.
- CBCT images suffer from motion artifacts caused by breathing and other movements.
- Current gating methods are limited to periodic motion, failing to address irregular movements.
Purpose of the Study:
- To develop a gate-less method for estimating and correcting both periodic and irregular motion in CBCT.
- To introduce CBCT-MOTUS, a novel technique for high temporal resolution motion and image reconstruction.
- To achieve per-projection temporal resolution of 182 ms for comprehensive motion correction.
Main Methods:
- A model-based, iterative reconstruction approach alternating between motion estimation and image correction.
- Motion estimation performed in projection space by comparing acquired and simulated projections.
- Utilizing B-spline parameterization, low-rank motion models, and spatial regularization for efficient motion field estimation.
Main Results:
- Successful estimation and correction of both periodic and irregular motion with high temporal resolution (182 ms).
- Demonstrated reduction of motion artifacts and deblurring on organ interfaces in corrected CBCT images.
- Validation across in silico, phantom, and in vivo clinical datasets.
Conclusions:
- Developed and validated CBCT-MOTUS, a gate-less, model-based method for comprehensive motion estimation and correction in CBCT.
- Achieved high temporal resolution by leveraging motion compressibility and smoothness with a low-rank B-spline model.
- Proof-of-principle results indicate potential for further clinical validation in radiotherapy.
