Thesis

160 Part II Chapter 6 Quantitative analysis of Na[18F]F-PET/CT: normal vs pathological bone To assess normal bone, in 16 patients the maximum of 5 bone measurements were performed, in 3 patients 4 VOI’s of normal bone were available and to quantify normal bone activity on Na[18F]F-PET/CT and in one patient a total of 3 VOIs. A correlation between SUL cut-off and LBM was demonstrated (R2=0.704, p=0.001; Spearman), which was clearly less strong between SUV cut-off and BW (R2=0.449, p=0.047; Spearman), pcritical=0.025. SUVs were equally distributed and SULs were differently distributed between genders (p=1.000 vs p=0.009, Mann Whitney U). The variation in SUV cut-off (median 8.7 g/mL; IQR 7.3-9.5 and total range 4.1-13.7 g/mL) was larger range than the predefined ± 10%, thus per-scan SUV cut-offs were used. Variation in incubation time could not explain SUV cut-off variability (R2= -0.169, p=0.477; Spearman). Very high interobserver agreement (ICC=0.964, p<0.001) was found for the SUV cut-off values with narrow limits of agreement in Bland-Altman analysis (mean difference: 0.14 g/mL, limits of agreement: -1.2 to +1.5 g/mL) without signs of heteroskedasticity (figure A). Figure A. Bland-Altman plot visualizing excellent interobserver agreement in SUV-cut-off values and no heteroscedasticity.

RkJQdWJsaXNoZXIy MjY0ODMw