Data-driven respiratory motion compensation for four-dimensional cone-beam computed tomography (4D-CBCT) using...

Area of Science:

• Medical Imaging
• Radiotherapy Physics
• Image Reconstruction

Background:

• Four-dimensional cone-beam CT (4D-CBCT) is crucial for image-guided radiotherapy.
• Respiratory motion during imaging degrades 4D-CBCT quality, introducing artifacts and reducing diagnostic accuracy.
• Current methods for motion compensation often rely on external surrogates or prior information.

Purpose of the Study:

• To assess the feasibility of a purely data-driven, a posteriori method for respiratory motion modeling and reconstruction compensation.
• To improve 4D-CBCT image quality under clinically relevant acquisition conditions.
• To evaluate the effectiveness of groupwise deformable image registration combined with motion-compensated reconstruction.

Main Methods:

• Evaluated groupwise deformable image registration and motion-compensated image reconstruction.
• Groupwise registration simultaneously aligns all temporal frames to a common reference.
• Applied motion compensation during Feldkamp-Davis-Kress (FDK) reconstruction using derived motion models.
• Assessed image quality via tissue interface sharpness (TIS), artifact mitigation, noise reduction, and contrast-to-noise ratio (CNR).

Main Results:

• Significant improvements in TIS recovery were observed, with mean-frame registration achieving near 100% recovery.
• Noise in soft tissue regions of interest was reduced by approximately 57-58%.
• Substantial average improvements in local CNR were noted, ranging from 93% to 116% depending on the method.

Conclusions:

• Data-driven groupwise registration and motion-compensated reconstruction are feasible for enhancing 4D-CBCT quality.
• These methods effectively reduce view aliasing artifacts and improve overall image fidelity.
• The approach offers a promising solution for improving image quality in clinical 4D-CBCT applications.