Liver and heart MR relaxometry in iron loading: reproducibility of three methods

Abstract

Purpose: To compare the derived T2* values and reproducibility of three methods used to assess iron-loading in heart and liver.

Materials and methods: In 23 pediatric patients, liver and cardiac gradient-echo imaging datasets (within-exam repeated sequence pairs) were evaluated. Data analyses compared derived relaxation values (average of pairs) and coefficient of variation (reproducibility of pairs).

Results: T2* values showed differences across methods, with pixel-wise mean > average fit > pixel-wise median. Coefficient of variation was found to be lower (better) with pixel-wise median and average fit methods compared to the pixel-wise mean technique. Maximum coefficient of variation values were lowest for the pixel-wise median approach in both the heart and liver.

Conclusion: Differences in derived T2* values between methods must be considered when comparing values to established magnetic resonance imaging (MRI)-biopsy formulas. The pixel-wise median and average fit methods demonstrate substantial benefits in reproducibility compared to the pixel-wise mean method. Since minimal variation in measurement is critical for patient care, median processing of relaxometry data may be preferable in both tissue types.

Keywords: MRI; T2*; heart; liver; relaxometry.