Cine cone beam CT reconstruction using low-rank matrix factorization: algorithm and a proof-of-principle study

Area of Science:

• Medical Imaging
• Computational Imaging
• Radiotherapy Physics

Background:

• Conventional 4D CBCT (4DCBCT) provides respiratory phase-resolved images, averaging patient anatomy over a breathing cycle.
• Clinical needs exist for true time-resolved CBCT (cine-CBCT) where each image represents an instantaneous projection.
• Reconstructing individual CBCT images from single projections is theoretically impossible due to information loss.

Purpose of the Study:

• To develop a computational model and algorithms for reconstructing semi-time-resolved cine-CBCT images.
• To explore the potential of exploiting sparsity, coherence, and redundancy within CBCT image sets.
• To enable more accurate representation of patient geometry's temporal variations.

Main Methods:

• Proposed a computational model based on low-rank approximation for cine-CBCT reconstruction.
• Utilized algorithms that exploit local and global temporal coherence in the patient anatomy.
• Leveraged shared redundant information across constituent CBCT images within a scan.

Main Results:

• Demonstrated the feasibility of reconstructing semi-time-resolved CBCT images (cine-CBCT).
• Showcased the ability to capture more temporal variation information compared to conventional 4DCBCT.
• Successfully exploited both local and global temporal coherence, including periodicity and slow variations.

Conclusions:

• The proposed low-rank approximation method enables the reconstruction of cine-CBCT, offering improved temporal resolution.
• This technique enhances the visualization of dynamic anatomical changes during respiration.
• Cine-CBCT holds promise for improving image guidance in radiotherapy and other clinical applications.