Figure 4:

Box and whisker plots of apparent R2* measured in all 97 biopsy samples compared with histologic iron grading, for, A, single-peak and, B, multipeak modeling of fat. Results show excellent correlation with iron grading. However, there were several outliers when single-peak modeling of fat was used. In most cases, outliers were caused by erroneous R2* measurements in the presence of high liver fat: patients 1 (95% fat), 2 (85% fat), 3 (80% fat [shown in Figure 1b]), and 4 (80% fat). Range of R2* values was different between the two reconstructions (particularly for iron groups 1 and 2) and was caused by overestimation of R2* in patients with high liver fat when single-peak modeling of fat was used.