Thoracic respiratory motion estimation from MRI using a statistical model and a 2-D image navigator

Area of Science:

• Medical Imaging
• Biomedical Engineering
• Computational Anatomy

Background:

• Respiratory motion significantly impacts medical imaging quality, particularly in PET-MR scans.
• Accurate modeling of intra- and inter-cycle breathing variability is crucial for effective motion correction.
• Existing respiratory motion models often struggle to capture the approximate repeatability of breathing patterns.

Purpose of the Study:

• To develop and describe a novel free-form nonrigid respiratory motion correction technique for the thorax.
• To create a motion model capable of capturing both intra- and inter-cycle respiratory motion variability.
• To enable real-time motion correction for simultaneous PET-MRI acquisition.

Main Methods:

• Principal Component Analysis (PCA) of motion states from dynamic 3D MRI data.
• Application of the motion model using a data-driven 2D MRI image navigator.
• Estimation of model applicability and determination of optimal image navigator positioning.

Main Results:

• The proposed technique successfully captures intra- and inter-cycle respiratory motion variability.
• Improvements of up to 40.5% in motion modeling accuracy were demonstrated compared to existing methods.
• The approach corrected up to 61% of the overall respiratory motion present in the data.
• Demonstrated application in MRI-based motion correction of real-time PET data.

Conclusions:

• The developed respiratory motion model effectively addresses intra- and inter-cycle variability.
• The technique offers significant improvements in motion correction accuracy for thorax imaging.
• This method holds promise for enhancing the quality of simultaneous PET-MRI acquisitions.