Empirical Motion-Artifact Reduction for Non-Rigid Motion in Dedicated Breast CT
IEEE Transactions on Bio-Medical Engineering
|April 21, 2025
Summary
This study introduces a new data-driven algorithm to reduce motion artifacts in breast CT scans. The method effectively minimizes artifacts caused by non-rigid patient motion without needing a motion model.
Area of Science:
- Medical Imaging
- Computational Imaging
Background:
- Dedicated breast CT offers 3D imaging without compression but is susceptible to motion artifacts due to slow gantry rotation.
- Non-rigid motion in breast CT is challenging to correct due to variable patient anatomy and movement patterns.
Purpose of the Study:
- To develop a data-driven empirical algorithm for reducing motion artifacts in dedicated breast CT.
- To address artifacts caused by non-rigid motion specific to breast imaging.
Main Methods:
- An iterative, data-driven empirical algorithm was developed for motion artifact reduction.
- The method utilizes b-spline fields in the image domain for transformations, updated via gradient descent and automatic differentiation.
- The algorithm operates without requiring an explicit model of the patient's motion.
Main Results:
- The algorithm was validated through simulation studies, physical phantoms, and clinical cases.
- Significant reduction in the appearance of motion artifacts was demonstrated across all tested scenarios.
Conclusions:
- A novel, fully data-driven empirical approach for motion artifact reduction in breast CT has been established.
- The developed algorithm effectively identifies and minimizes motion artifacts without relying on a predefined motion model, enhancing image quality.
