Motion correction and parameter estimation in dceMRI sequences: application to colorectal cancer

Area of Science:

• Medical Imaging
• Biophysics
• Computational Biology

Background:

• Dynamic contrast-enhanced MRI (dceMRI) is crucial for assessing tissue physiology.
• Non-rigid motion during dceMRI can significantly degrade image quality and complicate parameter estimation.
• Accurate motion correction is essential for reliable pharmacokinetic modeling.

Purpose of the Study:

• To develop a novel Bayesian framework for simultaneous non-rigid motion correction and pharmacokinetic parameter estimation in dceMRI.
• To integrate a physiological image formation model into the motion correction similarity measure.
• To improve the accuracy of physiological parameter estimation and image alignment in dceMRI.

Main Methods:

• A Bayesian framework was developed, maximizing the joint posterior probability of transformations and physiological parameters.
• The similarity measure incorporated a physiological image formation model.
• Non-rigid motion correction utilized the diffeomorphic logDemons algorithm.
• The framework was validated using simulated and real dceMRI datasets.

Main Results:

• The proposed method demonstrated improved co-registration of dceMRI images.
• Enhanced accuracy in the estimation of pharmacokinetic parameters was observed.
• The joint optimization approach effectively addressed motion artifacts and parameter estimation.

Conclusions:

• The novel Bayesian framework offers a robust solution for motion correction in dceMRI.
• Simultaneous motion correction and parameter estimation improve quantitative analysis of dceMRI data.
• This method holds potential for advancing diagnostic capabilities using dceMRI.