A method to map errors in the deformable registration of 4DCT images

Area of Science:

• Medical Imaging
• Computational Anatomy
• Radiotherapy Physics

Background:

• Deformable image registration (DIR) is crucial for analyzing anatomical changes in 4D CT datasets.
• Accurate error estimation in DIR is essential for reliable quantitative analysis and treatment planning.
• Current validation methods using landmarks may not capture the full spatial extent of registration uncertainties.

Purpose of the Study:

• To develop and present a novel computational approach for estimating errors in deformable image registration (DIR) applied to 4D CT data.
• To address the limitations of point-based validation by mapping the complete spatial distribution of DIR errors.

Main Methods:

• Generated displacement vector fields (DVFs) from sequential 4D CT phases.
• Utilized principal component analysis (PCA) on DVFs and physical landmark data to differentiate anatomical motion from registration errors.
• Reconstructed DIR error maps by isolating error eigenmodes.

Main Results:

• Simulated DVFs exhibited principal component properties similar to those in actual 4D CT data.
• The proposed method accurately recovered simulated DIR error maps.
• Demonstrated the method's efficacy using a numerical model incorporating breathing motion and simulated spatially correlated DIR errors.

Conclusions:

• DIR errors can exhibit complex spatial patterns, challenging traditional landmark-based validation.
• The developed method allows for comprehensive mapping of DIR errors, improving uncertainty quantification.
• This technique offers a more representative assessment of registration uncertainties beyond discrete validation points.