Sparse 3D Radon Space Rigid Registration of CT Scans: Method and Validation Study

Area of Science:

• Medical Imaging
• Image Registration
• Computational Imaging

Background:

• Computed tomography (CT) registration is crucial for medical image analysis.
• Current methods often require dense sampling, leading to high radiation exposure.
• Efficient registration with reduced dose is highly desirable.

Purpose of the Study:

• To develop a novel rigid registration method for 3D CT datasets in Radon space.
• To utilize sparse sampling of scanning projections for efficient registration.
• To reduce X-ray dose while maintaining registration accuracy.

Main Methods:

• The method operates in 3D Radon space, using 3D Radon transforms of CT scans.
• It matches sparse projection data to dense projection data.
• Registration is achieved by solving linear equations or non-linear optimization based on projection direction vectors.

Main Results:

• The proposed method accurately determines rigid transformations between CT datasets.
• It outperforms image space registration in convergence range for 3D parallel beam data.
• Significantly reduced X-ray dose is achieved compared to full CT scans.

Conclusions:

• Radon space registration with sparse sampling offers an efficient and low-dose alternative for CT image registration.
• The method shows promise for improving CT imaging protocols.
• Further validation across different CT geometries (fan-beam, cone-beam) is supported.