Improved estimation of MR relaxation parameters using complex-valued data.

Area of Science:

• Magnetic Resonance Imaging (MRI)
• Medical Physics
• Image Analysis

Background:

• Traditional MR image analysis often uses magnitude data for T1, T2, and T2* maps.
• Low signal-to-noise ratio (SNR) in magnitude-based methods leads to biased parameter estimates.
• This bias confounds comparisons across different SNRs or MRI scanners.

Purpose of the Study:

• To compare various estimation techniques using real or complex-valued MR data.
• To achieve unbiased estimation of MR relaxation parameters without additional preprocessing.
• To address limitations of magnitude-based methods in low SNR conditions.

Main Methods:

• Comparison of existing and novel complex-valued data techniques against standard magnitude-based methods.
• Evaluation of bias, variance, and processing time via simulations across parameter variations.
• Validation using experimental measurements with degraded SNR.

Main Results:

• Complex-valued data techniques demonstrated superior performance over magnitude-based methods, even those accounting for Rician noise.
• These methods achieved statistically efficient, minimum variance unbiased estimates.
• Computational costs were comparable or slightly higher (≈2x) than magnitude-based estimators.

Conclusions:

• Complex-valued data estimation techniques offer superior, unbiased parametric maps compared to magnitude-based methods.
• These advanced techniques also provide phase and field maps, unavailable with traditional approaches.
• The findings advocate for the adoption of complex-valued data analysis for more robust MR parameter mapping.