Adapting population liver motion models for individualized online image-guided therapy

Area of Science:

• Medical Physics
• Radiotherapy
• Image Registration

Background:

• Respiratory motion in abdominal cancer patients causes daily variations, impacting radiotherapy delivery.
• Current deformable registration methods for motion analysis are computationally intensive and require user intervention, limiting online application.

Purpose of the Study:

• To develop a rapid technique for quantifying patient breathing motion from volumetric images.
• To adapt a population liver respiratory motion model for online radiotherapy adjustments.

Main Methods:

• A novel method quantifies 1D shifts in image intensities within navigator channels on respiratory-sorted volumetric images.
• The 1D motion data from liver edges was used to adapt a population liver respiratory motion model.
• Validation involved deformable registration (MORFEUS) and landmark-based analysis (vessel bifurcations).

Main Results:

• The adapted model accurately described patient respiratory motion in superior-inferior (SI) and anterior-posterior (AP) directions (0.26 ± 0.11 cm and 0.30 ± 0.21 cm).
• Tumor center of mass (COM) motion prediction accuracy was 0.30 ± 0.22 cm and 0.34 ± 0.31 cm.
• Absolute validation using bifurcations showed high accuracy (0.26 ± 0.16 cm and 0.13 ± 0.04 cm in SI and AP directions).

Conclusions:

• The developed technique enables rapid quantification and adaptation of respiratory motion models for online radiotherapy.
• This method complements existing biomechanical registration techniques, enhancing real-time treatment delivery for abdominal cancer patients.