ESTRO 2025 - Abstract Book

S2955

Physics - Image acquisition and processing

ESTRO 2025

894

Proffered Paper CBCT-MOTUS: motion correction in CBCT imaging using gate-less model-based reconstruction of non-rigid motion and images Ethan Waterink 1 , Rodrigo José Santo 1 , Cornelis A. T. van den Berg 1 , Hugo W.A.M. de Jong 1,2 , Alessandro Sbrizzi 1 , Casper Beijst 1 1 Radiotherapy, UMC Utrecht, Utrecht, Netherlands. 2 Radiology, UMC Utrecht, Utrecht, Netherlands Purpose/Objective: Cone beam computed tomography (CBCT) plays a crucial role in verifying patient positioning during radiotherapy. However, CBCT images are often blurred by motion such as breathing, bowel motion and/or repositioning. Conventional methods employ gating techniques to mitigate motion artifacts by assuming periodicity which is restricted to respiratory motion. However, this does not solve irregular, a-periodic motion [1]. We present a novel method for gate-less joint reconstruction of motion and images, thereby estimating and correcting for all non-rigid motion at high temporal frequency (182ms). Material/Methods: To this end, we have developed a framework (Figure 1) for gate-less joint non-rigid motion estimation and image correction (CBCT-MOTUS) from reference images and projection data, to estimate irregular and periodic 3D motion. Starting from a non-corrected motion-blurred image reconstruction, we perform the reconstruction process by alternating between motion estimation and correction. Motion estimation is performed directly in projection space by comparing acquired projections to simulated projections that take motion-fields into account. The optimization process benefits from assumptions and specific requirements including (i) compressing the motion-fields using B spline parameterization to reduce the number of parameters to estimate, (ii) exploiting spatiotemporal correlation of motion using a low-rank motion model [2], and (iii) enforcing smoothness using spatial regularization to include a priori knowledge on motion-fields. High temporal resolution (182 ms for this scanner) is achieved as motion-fields are estimated for each acquired gantry angle. It is validated in-silico and on phantom acquisitions, and applied on clinical acquisitions.

Results: The framework can estimate and correct irregular and periodic motion with high temporal resolution (182 ms), as shown in in-silico simulations, phantom acquisitions and clinical data, most notably for patient 1's irregular breathing (Figure 2). The motion-corrected reconstructed images show improved image features for all experiments, such as deblurring. Interfaces regain their sharpness, and patient 2's tumor is more clearly defined.