Patient-bounded extrapolation using low-dose priors for volume-of-interest imaging in C-arm CT.

Area of Science:

• Medical Imaging
• Radiology
• Image Reconstruction

Background:

• C-arm systems enable low-dose 3D volume-of-interest (VOI) imaging.
• Lateral truncation in projections causes artifacts in conventional reconstruction.
• Existing extrapolation methods often rely on ad-hoc assumptions.

Purpose of the Study:

• To improve 3D VOI image quality by leveraging patient-specific prior information.
• To address severe truncation artifacts in C-arm based VOI imaging.
• To integrate a new extrapolation method into existing workflows.

Main Methods:

• A patient-bounded extrapolation method using pre-acquisition fluoroscopic images.
• Extraction of 2D patient contours from AP and ML views.
• Estimation of a 3D patient shape model to define projection boundaries.

Main Results:

• The proposed method achieved a relative root mean square error (rRMSE) of ~1.0% on clinical datasets.
• This represents a significant improvement over heuristic extrapolation techniques (rRMSE of 8.0%).
• Effective reduction of truncation artifacts even with severe data truncation.

Conclusions:

• The novel method substantially enhances image quality in 3D C-arm VOI imaging.
• No additional radiation is required as it utilizes existing fluoroscopic images.
• The technique integrates seamlessly into current interventional workflows without extra hardware.