Optimized bias and signal inference in diffusion-weighted image analysis (OBSIDIAN).

Area of Science:

• Medical Imaging
• Signal Processing

Background:

• Rician bias in magnitude MR images affects signal accuracy.
• Accurate noise estimation is crucial for quantitative MRI analysis.

Purpose of the Study:

• To develop and evaluate OBSIDIAN, a novel method for Rician bias correction in magnitude MR images.
• To assess the performance of OBSIDIAN compared to maximum likelihood estimation (MLE).

Main Methods:

• A model-based, iterative fitting procedure estimates true signal and Gaussian noise pixel-by-pixel.
• A precomputed function relates residuals to noise standard deviation for iterative estimation.
• The method was tested using diffusion signal decay simulations and diffusion-weighted prostate MRI data.

Main Results:

• OBSIDIAN demonstrates speed and accuracy comparable to MLE in simulations with pure Gaussian noise.
• Composite fitting improves parameter prediction accuracy in low signal-to-noise ratio (SNR) scenarios.
• Good agreement was observed between OBSIDIAN and high SNR reference data for prostate diffusion.

Conclusions:

• OBSIDIAN offers a rapid and simple approach for Rician bias correction.
• The composite fitting approach enables accurate parameter estimation in low SNR clinical settings.
• This method simplifies the comparison of diffusion parameters across studies.