An MRI framework for respiratory motion modelling validation.

Area of Science:

• Medical Imaging
• Radiotherapy Physics
• Computational Anatomy

Background:

• Respiratory motion models link 4D imaging to surrogates for estimating 3D breathing motion.
• Evaluating these models on real patient data requires a robust validation framework.
• Magnetic Resonance Imaging (MRI) offers a suitable modality for such evaluations.

Purpose of the Study:

• To propose and validate a novel MRI-based framework for evaluating respiratory motion models.
• To assess the performance of these models using real patient data with intra- and inter-fraction variability.
• To determine the accuracy of time-resolved (TR) 3D breathing motion estimation.

Main Methods:

• Utilized respiratory-correlated (RC) 4D MRI to build motion models.
• Employed deformable image registration (DIR) for model building and inter-fraction variation compensation.
• Validated models using both TR 2D cine-MRI and additional 4DMRI data for 2D and 3D assessments.

Main Results:

• Deformable image registration (DIR) error was comparable to voxel size, increasing with inter-fraction motion.
• 2D validation showed median estimation error below 4 mm; 3D validation yielded 1.33 mm (intra-fraction) and 4.21 mm (inter-fraction).
• Model performance correlated with DIR error, with reduced accuracy during inter-fraction baseline variations.

Conclusions:

• The proposed MRI framework shows potential for evaluating global motion models.
• This framework is valuable for organ motion management in MRI-guided radiotherapy.
• Accurate assessment of breathing motion variability is key for treatment precision.