Multisite multivendor validation of a quantitative MRI and CT compatible fat phantom.

Area of Science:

• Medical Imaging
• Radiology
• Biomedical Engineering

Background:

• Chemical shift-encoded MRI accurately quantifies liver fat using proton density fat-fraction (PDFF).
• Computed tomography (CT) quantifies fat by measuring attenuation, but current phantoms do not accurately represent human liver CT numbers.
• Accurate phantoms are crucial for validating and harmonizing quantitative imaging techniques.

Purpose of the Study:

• To develop and validate an optimized phantom that simultaneously mimics MRI and CT signals of fatty liver.
• To create a tool for accurate, cross-platform liver fat quantification.

Main Methods:

• An agar-based phantom with 12 vials of varying fat fractions and iodinated contrast was designed in a spherical housing.
• A multi-site, multi-vendor validation study was conducted using 1.5T and 3T MRI and CT scanners.
• Image analysis involved measuring MRI-PDFF and CT numbers (HU) within regions of interest, with statistical analysis using mixed-effects models, linear regression, and Bland-Altman analysis.

Main Results:

• MRI-PDFF measurements closely matched nominal fat fractions across field strengths and vendors.
• A strong linear relationship (R²=0.999) was observed between MRI-PDFF and CT numbers, reflecting in vivo behavior.
• Excellent test-retest repeatability and good reproducibility were demonstrated for both MRI and CT measurements across vendors.

Conclusions:

• The developed phantom successfully mimics quantitative liver fat signals for both MRI and CT.
• This phantom can promote wider adoption and standardization of liver fat quantification methods across institutions, vendors, and imaging platforms.