Fat-water separation based on Transition REgion Extraction (TREE).

Area of Science:

• Medical Imaging
• Biomedical Engineering
• Magnetic Resonance Imaging

Background:

• Accurate fat-water separation is crucial for quantitative MRI.
• Existing methods struggle with low signal-to-noise ratio (SNR), B0 field inhomogeneity, and complex anatomies.

Purpose of the Study:

• To develop a robust fat-water separation and quantification method using fat-water transition region extraction (TREE).
• To address challenges like low SNR, fast-changing B0 fields, and disjointed anatomies.

Main Methods:

• The TREE method categorizes pixel phasor solutions into fat-dominant and water-dominant groups.
• It extracts the fat-water transition region by detecting phasor map changes.
• Smoothest phasor combinations are used for transition region pixels, with surrounding pixels informing other regions.

Main Results:

• The TREE method achieved a quantitative score comparable to the winner of the ISMRM 2012 Challenge.
• It demonstrated robustness in challenging scenarios, including low SNR (~10) and fast-changing B0 fields.
• Accurate proton density fat fraction results were obtained without observed fat-water swap.

Conclusions:

• A novel fat-water transition region extraction (TREE) method provides robust water-fat separation and quantification.
• The TREE method is effective in challenging imaging conditions such as spatially disjointed objects, B0 field variations, and low SNR.