Adapting liver motion models using a navigator channel technique.

Area of Science:

• Medical Physics
• Image-guided therapy
• Computational anatomy

Background:

• Deformable registration is crucial for accurate tumor targeting in radiation therapy.
• Online applications require both efficiency and accuracy in deformable registration.
• Existing methods may lack the speed needed for real-time image-guided interventions.

Purpose of the Study:

• To develop a fast deformable registration technique for image-guided radiation therapy.
• To combine biomechanical and patient-specific motion models for efficient liver motion prediction.
• To improve the accuracy of tumor targeting through real-time motion assessment.

Main Methods:

• Developed a navigator channel technique integrating a biomechanical model-based deformable registration algorithm with population and patient-specific motion models.
• Generated a respiratory population-based liver motion model using finite element modeling from CT data.
• Adapted the population model to individual patient motion using a navigator channel on liver boundaries from exhale and inhale CT images.

Main Results:

• The population liver motion model showed average displacements of 0.12 cm (LR), 0.84 cm (AP), and 1.24 cm (SI).
• The navigator channel accurately predicted 1D SI and AP liver motion (<0.11 cm), below the voxel size of 0.25 cm.
• Verification using vessel bifurcation, tumor center of mass, and MORFEUS showed residual errors on the order of voxel image sizes.

Conclusions:

• The navigator channel technique provides fast and accurate deformable registration for image-guided radiation therapy.
• This method enables online assessment and monitoring of intrafraction motion.
• Potential to significantly improve tumor targeting accuracy and treatment efficacy in radiation therapy.