MR-MOTUS: model-based non-rigid motion estimation for MR-guided radiotherapy using a reference image and minimal...

Area of Science:

• Medical Imaging
• Computational Physics
• Radiotherapy Physics

Background:

• The integration of MRI with linear accelerators (MR-Linac) necessitates accurate, time-resolved estimation of internal organ motion from MRI data for adaptive radiation therapy.
• Current methods for time-resolved motion estimation from MRI data present significant challenges, limiting real-time applications.

Purpose of the Study:

• To introduce MR-MOTUS, a framework designed for efficient, non-rigid 3D motion estimation from limited k-space MRI data.
• To enable rapid and precise motion tracking for adaptive radiotherapy planning and delivery in MR-Linac systems.

Main Methods:

• MR-MOTUS utilizes a signal model linking k-space signals to non-rigid motion fields and a reference image.
• It employs model-based reconstruction of motion fields directly from undersampled k-space data, leveraging spatial correlation of internal body motion.
• Motion fields are represented in a low-dimensional space for rapid reconstruction from minimal data.

Main Results:

• MR-MOTUS successfully reconstructed in vivo 3D rigid head motion from highly undersampled (474-fold) k-space data.
• It also reconstructed in vivo non-rigid 3D respiratory motion from significantly undersampled (63-fold) k-space data.
• Preliminary results with prospectively undersampled data confirm the method's practical feasibility for real-time applications.

Conclusions:

• MR-MOTUS provides a robust framework for time-resolved, non-rigid 3D motion estimation from minimal MRI data.
• The method holds significant potential for enhancing adaptive radiotherapy delivery in MR-Linac settings by enabling real-time motion compensation.
• Further validation on prospectively acquired data supports its clinical translation for improved patient outcomes.