Detecting patient motion in projection space for cone-beam computed tomography

Area of Science:

• Medical Imaging
• Radiology
• Image Reconstruction

Background:

• Cone-beam X-ray systems require stationary objects for high-quality 3D reconstructions.
• Patient motion during acquisition significantly degrades image quality and diagnostic accuracy.
• Existing methods struggle to isolate patient motion from the inherent rotational movement of the imaging system.

Purpose of the Study:

• To develop a novel similarity metric capable of distinguishing patient motion from system rotation in cone-beam X-ray projections.
• To enable the modeling and compensation of patient motion directly from projection data.
• To improve the accuracy and reliability of 3D reconstructions in cone-beam CT.

Main Methods:

• A new similarity metric was designed to analyze successive X-ray projection images.
• The metric was developed to differentiate between the expected rotational motion of the source-detector assembly and unintended patient motion.
• Quantitative evaluation was performed using simulated dental cone-beam X-ray data.

Main Results:

• The proposed similarity metric successfully distinguished between rotational and additional patient motion.
• Quantitative analysis on simulated data demonstrated the metric's effectiveness.
• Qualitative assessment using real patient data confirmed the metric's practical performance.

Conclusions:

• The novel similarity metric offers a promising approach for identifying and potentially correcting patient motion artifacts in cone-beam X-ray imaging.
• This method facilitates motion-aware 3D reconstruction, leading to improved image quality.
• The findings have significant implications for enhancing diagnostic accuracy in dental and other cone-beam CT applications.