How do registration parameters affect quantitation of lung kinematics?

Area of Science:

• Pulmonary imaging and biomechanics
• Medical physics and computational anatomy

Background:

• Quantifying lung motion is crucial for medical applications but remains a significant challenge.
• Accurate assessment of respiratory kinematics is essential for improving treatments and diagnostics.

Purpose of the Study:

• To evaluate the accuracy of motion estimation techniques for isolated lungs.
• To compare different registration metrics and image resolutions for assessing respiratory kinematics.

Main Methods:

• Utilized four-dimensional computed tomography (4DCT) on isolated porcine lungs in an artificial thorax.
• Employed pairwise non-rigid registration with various metrics (mutual information, cross-correlation, mean squares) and resolutions.
• Validated accuracy using manually identified landmarks and known displacements.

Main Results:

• Motion quantitation precision decreases significantly with increasing inflation intervals between images.
• The impact of image resolution on motion estimation sensitivity is anatomically dependent.
• Mutual information and cross-correlation metrics performed similarly and were superior to mean squares; however, none overcame challenges with large inflation intervals.

Conclusions:

• The findings provide insights into optimizing the quantification of pulmonary kinematics.
• Results will guide more effective and efficient motion estimation in future research and clinical applications.
• Further exploration of parameter combinations is warranted to enhance accuracy.