Motion-map constrained image reconstruction (MCIR): application to four-dimensional cone-beam computed tomography.

Area of Science:

• Medical Imaging
• Radiation Oncology
• Image Reconstruction

Background:

• Respiratory correlated four-dimensional cone-beam computed tomography (4DCBCT) is crucial for verifying internal target motion and volume before treatment.
• Standard CBCT reconstruction for 4DCBCT faces challenges due to insufficient projection data, leading to streaking artifacts.

Purpose of the Study:

• To develop a novel 4DCBCT reconstruction algorithm, motion-map constrained image reconstruction (MCIR).
• To achieve high-quality, high phase-resolution 4DCBCT images with standard imaging dose and projection data.

Main Methods:

• Modeled 4DCBCT as a combination of free-breathing 3DCBCT (FB-3DCBCT) and phase-specific update vectors using a motion-map matrix.
• Employed compressed sensing (CS) reconstruction to update moving voxels aggressively and stationary voxels minimally.
• Evaluated MCIR using numerical phantoms and a lung cancer patient, comparing with FDK and PICCS methods.

Main Results:

• MCIR successfully reconstructed high phase-resolved 4DCBCT (up to 20 phases) from typical FB-3DCBCT scan data without compromising image quality.
• MCIR demonstrated excellent image quality compared to other published algorithms.
• Reconstruction achieved without streak artifacts caused by insufficient projections.

Conclusions:

• MCIR offers a potential solution for high-quality 4DCBCT in image-guided radiation therapy (IGRT) without increased imaging dose.
• The algorithm enables reconstruction of at least 20 phase images using standard FB-3DCBCT projection data, eliminating artifacts.