MRI for assessing abdominal organs iron concentration: A comparative study between the relaxometry methods

Objective

To compare magnetic resonance imaging (MRI) commercial 3D quantitative Dixon sequence (qDixon) and 2D multi-gradient recalled echo sequence (GRE) for iron quantification in multiple abdominal organs.

Methods

1.5T MRI GRE and qDixon data were collected on patients with 211 MR exams from 171 patients (86 males, 85 females; median age: 21 years). Compare the R2* values of liver, pancreas, spleen, and kidneys using Bland-Altman, intraclass correlation coefficients (ICC), concordance correlation coefficients (CCC), and linear regression. Iron overload (IO) diagnostic concordance was assessed using overall agreement and the Kappa coefficient.

Results

Bland-Altman analysis of liver, pancreas, spleen, and kidney R2* values between GRE and qDixon, resulted in a bias (absolute mean difference) of −11.3 1/s (LoA: 77.7 and −100.3), −11.2 1/s (LoA: 123.9 and −146.3), 5.1 1/s (LoA: 117.3 and −107.1), and 1.9 1/s (LoA: 14.7 and −10.9). The CCCs between GRE and qDixon R2* values for liver, pancreas, spleen, and kidneys were 0.98, 0.94, 0.96, and 0.95, the ICCs were 0.99, 0.95, 0.96, and 0.97, respectively. Linear regression analysis correlating abdominal organs R2* values of GRE and qDixon resulted in a coefficient of determination of 0.96, 0.89, 0.93, and 0.92 (all P < 0.001). The overall agreement was 98.5 %, 94.8 %, 92.6 %, and 90.7 %; the Kappa value was 0.95, 0.89, 0.84, and 0.69 (all P < 0.001).

Conclusion

The qDixon and GRE showed good agreement and significant positive in measuring R2* values for IO assessment in abdominal organs, with qDixon being an excellent adjunctive diagnostic method.