Reconstruction of undersampled dynamic images by modeling the motion of object elements

Area of Science:

• Medical Imaging
• Biomedical Engineering
• Image Reconstruction

Background:

• Dynamic MRI acquisition is time-limited due to data requirements for image sequence reconstruction.
• Existing undersampled reconstruction methods often model temporal information using pixel intensity variations.
• These methods face limitations in capturing complex dynamic processes efficiently.

Purpose of the Study:

• To develop a novel undersampled dynamic MRI reconstruction technique.
• To improve image quality and reduce acquisition time in dynamic MRI.
• To enable quantification of dynamic information through motion estimation.

Main Methods:

• Proposed a new dynamic MRI reconstruction approach based on motion estimation of image object elements (obels).
• Assumed that obel displacement has lower bandwidth than pixel intensity fluctuations for efficient modeling.
• Utilized fewer parameters to model motion compared to acquired samples.

Main Results:

• Successfully reconstructed cardiac and joint images with undersampling factors of 8 and 4, respectively.
• Achieved root mean square (RMS) errors below 4% in preliminary results.
• Obtained approximate motion vectors for each obel during the reconstruction process.

Conclusions:

• The proposed motion-based reconstruction method effectively addresses dynamic MRI limitations.
• This technique allows for accurate dynamic MRI reconstruction and quantification of motion.
• It is suitable for dynamic studies where obel motion models are parameter-efficient.