Bayesian estimation of changes in transverse relaxation rates

Area of Science:

• Magnetic Resonance Imaging (MRI)
• Biophysics
• Medical Imaging Analysis

Background:

• R(2)* estimation in MRI is susceptible to bias when assuming normally distributed magnitude data.
• The impact of this assumption on changes in R(2)* (DeltaR(2)*), particularly after contrast agent administration, remains largely uninvestigated.

Purpose of the Study:

• To develop and evaluate novel Bayesian maximum a posteriori (MAP) approaches for robust voxelwise DeltaR(2)* estimation.
• To compare DeltaR(2)* estimation using normally distributed versus Rice-distributed MRI magnitude data.
• To quantify the uncertainty in DeltaR(2)* estimates and assess significant signal enhancement probability.

Main Methods:

• Implementation of two Bayesian MAP algorithms for DeltaR(2)* estimation: one assuming normal data distribution, the other assuming Rice distribution.
• In vivo evaluation of the developed techniques using ultrasmall superparamagnetic iron oxide particles (USPIOs) in orthotopic murine prostate tumors.
• Assessment of the biasing effects of the normality assumption on DeltaR(2)* under varying signal-to-noise ratios (SNRs).

Main Results:

• The assumption of normally distributed MRI magnitude data leads to underestimation of DeltaR(2)* at modest signal-to-noise ratios.
• The observed underestimation bias in DeltaR(2)* is less pronounced compared to the bias in R(2)* estimates.
• The Bayesian approach provides robust, voxelwise uncertainty quantification for DeltaR(2)*, enabling reliable assessment of enhancement probability.

Conclusions:

• The novel Bayesian MAP approach offers a statistically sound framework for accurate DeltaR(2)* quantification in MRI.
• The assumption of normally distributed noise is more justifiable for DeltaR(2)* analysis than for precontrast R(2)* estimation.
• This technique enhances the reliability of detecting signal changes induced by contrast agents in preclinical MRI studies.