Choreography Controlled (ChoCo) brain MRI artifact generation for labeled motion-corrupted datasets.

Area of Science:

• Medical Imaging
• Biomedical Engineering
• Neuroscience

Background:

• Patient motion is a significant limitation in Magnetic Resonance Imaging (MRI) acquisition.
• Retrospective motion correction techniques require labeled, motion-corrupted datasets for validation, which are currently scarce.
• Existing methods lack standardized protocols for generating reproducible motion-corrupted MRI data.

Purpose of the Study:

• To propose a novel, straightforward, and reproducible methodology for acquiring motion-corrupted MRI images.
• To validate the performance of motion estimation techniques using controlled head movements.
• To facilitate the development and benchmarking of retrospective MRI motion correction algorithms.

Main Methods:

• Developed an MRI-compatible system with a visual target and customized glasses for controlled head choreographies.
• Utilized rigid-body volume registration of fast 3D echo-planar imaging (EPI) time series for motion estimation.
• Implemented a spatio-temporal upsampling and interpolation method to handle fast motion.

Main Results:

• Demonstrated accurate head position determination with an average standard deviation of approximately 0.39 degrees.
• The proposed protocol allows for reproducible generation of motion-corrupted MRI data.
• The methodology is compatible with all MRI systems and provides insights into motion artifact origins.

Conclusions:

• The presented protocol offers a versatile solution for creating labeled, motion-corrupted MRI datasets.
• This work supports the MRI and artificial intelligence research communities in developing and testing motion correction algorithms.
• The methodology can enhance the reliability and accuracy of MRI scans affected by patient movement.