A Feature-based Affine Registration Method for Capturing Background Lung Tissue Deformation for Ground Glass Nodule...

Area of Science:

• Medical Imaging
• Radiology
• Computational Anatomy

Background:

• Lung nodule assessment using computed tomography (CT) relies on size measurements.
• Lung tissue deformation between scans can distort nodule size, leading to inaccurate disease progression assessment.
• Accurate nodule tracking requires compensation for background lung deformation.

Purpose of the Study:

• To propose and evaluate a feature-based affine registration method for lung nodule tracking.
• To compare its performance against other registration techniques in compensating for lung deformation.
• To ensure reliable assessment of disease-induced nodule changes.

Main Methods:

• Implemented and tested a feature-based affine registration method, alongside deformable and least-square fit methods.
• Utilized ten patient CT datasets with twelve nodules (GGNs, part-solid, solid).
• Evaluated registration accuracy using target registration error (TRE) across 30-50 fiducial landmarks per lesion.

Main Results:

• The proposed feature-based affine lesion-centered registration achieved a TRE of 1.1 ± 1.2 mm.
• Deformable registration yielded a slightly better TRE (1.2 ± 1.2 mm) but is computationally intensive.
• Feature-based affine registration is computationally efficient and avoids issues associated with deformable methods.

Conclusions:

• Feature-based affine registration effectively compensates for lung tissue deformation in nodule tracking.
• This method provides a reliable baseline for assessing lung nodule changes due to disease.
• It offers a more robust alternative to methods relying on ambiguous nodule segmentation.