Technical note: a physical phantom for assessment of accuracy of deformable alignment algorithms

Area of Science:

• Medical Imaging
• Image Processing
• Computational Anatomy

Background:

• Deformable image registration (DIR) is crucial for medical image analysis, particularly in radiotherapy and surgical guidance.
• Accurate validation of DIR algorithms is essential for clinical translation and reliable outcomes.
• Current validation methods often lack ground truth or are signal-dependent, limiting their scope.

Purpose of the Study:

• To investigate the feasibility of a simple deformable phantom for quality assurance (QA) of DIR algorithms.
• To establish a ground truth deformation field for quantitative comparison of DIR methods.
• To assess the phantom's capability in evaluating DIR accuracy independently of imaging signal.

Main Methods:

• A diagnostic thoracic imaging phantom with a deformable foam insert containing plastic markers was developed.
• Markers formed a lattice to provide measurable deformation data under superior-inferior compression simulating breathing.
• Marker positions were manually identified to create a known deformation field; digital removal preceded registration testing.

Main Results:

• Repeat measurements of marker positions demonstrated an accuracy better than 1 mm.
• Testing with multiple DIR algorithms confirmed the system's ability to quantitatively evaluate registration errors.
• The phantom provided a deformation measure independent of the signals driving the deformation parameters.

Conclusions:

• A simple deformable phantom is feasible and effective for QA of DIR algorithms.
• This phantom enables quantitative assessment of DIR accuracy using a signal-independent metric.
• The developed phantom serves as a valuable tool for validating DIR algorithm performance in medical imaging.