Optimized raw data selection for artifact reduction of breathing controlled four-dimensional sequence scanning.

Area of Science:

• Medical Imaging
• Radiotherapy Physics
• Image Reconstruction

Background:

• Four-dimensional computed tomography (4DCT) is crucial for radiotherapy, but artifacts from breathing variations degrade image quality.
• Existing breathing-controlled 4DCT algorithms struggle with artifacts caused by significant breathing changes, impacting treatment accuracy.
• These artifacts can lead to negative consequences in radiotherapy planning and delivery.

Purpose of the Study:

• To introduce and evaluate a novel raw data selection and binning algorithm for 4DCT.
• To improve image quality and geometric accuracy in 4DCT imaging.
• To validate the algorithm's performance using phantom studies without additional radiation exposure.

Main Methods:

• Phantom examinations were conducted to assess the new optimized raw data selection and binning algorithm.
• Geometric accuracy was evaluated using metrics like volume fidelity, root mean square error, and Dice coefficient.
• Tumor motion trajectories included one-dimensional, three-dimensional, and motion hysteresis effects, with scans without longer breathing cycles as references.

Main Results:

• The optimized algorithm achieved a maximum median volume deviation of 1% compared to reference scans across all movements.
• Standard reconstruction methods showed significantly higher median deviations (9%–21%) for various motion types.
• The optimized algorithm yielded high median Dice coefficients (0.918–0.975) and reduced median root mean square errors (30–120 HU²) compared to standard reconstruction.

Conclusions:

• The developed algorithm effectively reduces 4DCT artifacts stemming from missing projection data.
• The approach enhances image quality and geometric accuracy without increasing patient radiation dose.
• Improved 4DCT image quality is expected to benefit radiotherapy treatment planning.