Maximum Entropy Approach in Dynamic Contrast-Enhanced Magnetic Resonance Imaging

Area of Science:

• Medical Imaging
• Biophysics
• Computational Biology

Background:

• Accurate estimation of physiological kinetic parameters in Dynamic Contrast-Enhanced Magnetic Resonance Imaging (DCE-MRI) relies heavily on determining the arterial input function (AIF).
• Challenges arise in DCE-MRI analysis when direct measurement of the AIF is not feasible.

Purpose of the Study:

• To propose a novel Bayesian method for estimating DCE-MRI physiological parameters and the AIF simultaneously, particularly in scenarios lacking AIF measurements.
• To validate the proposed method's ability to reliably determine the AIF and subsequently estimate kinetic parameters.

Main Methods:

• A Bayesian approach combining the Maximum Entropy Method (MEM) for AIF prior probability distribution specification and the Maximum A Posteriori (MAP) approach for kinetic parameter estimation.
• Validation of AIF estimation accuracy using Kullback-Leibler divergence.
• Evaluation of the algorithm using a breast cancer DCE-MRI dataset.

Main Results:

• The proposed MEM-MAP algorithm reliably determines the arterial input function (AIF) directly from DCE-MRI data.
• Subsequent estimation of physiological kinetic parameters is enabled with improved accuracy.

Conclusions:

• The Maximum Entropy Method (MEM) provides a robust tool for reconstructing probability distributions and assessing the input function from existing DCE-MRI data.
• This approach offers an alternative for AIF assessment, leading to better data fitting and more reliable estimation of kinetic parameters, ultimately enhancing the utility of DCE-MRI.