Intra-voxel incoherent motion parameters of rectal cancer at 3.0 T MRI: effect of region of interest on measurement consistency of parameters

Objective: To determine the measurement consistency of diffusion coefficient D, perfusion fraction f and pseudodiffusion coefficient D^* in rectal cancers based on different ROIs. Methods: Forty-three patients with histologically proven rectal cancers were examined using echo-planar DW-MRI with eight b values (0 to 1000 s/mm²). Intravoxel incoherent motion parameters were measured on intravoxel incoherent motion map that contained the largest tumor cross-section, according to two distinct ROI protocols: freehand outline ROI and semi-automatic tumor center ROI. The two protocols were compared for differences in IVIM parameters and the interclass correlation coefficient (ICC) were also calculated. intra-and inter-observer variability using paired t test and Bland-Altman plot. Results: The IVIM parameters(D, f and D^*) obtained by ROIs for outlined and center analysis were (1.08 ± 0.24) × 10^-3 mm²/s, (0.16 ± 0.06), (26.59 ± 19.54) × 10^-3 mm²/s and (1.06 ± 0.27) × 10^-3 mm²/s, (0.17 ± 0.07), (30.79 ± 20.85) × 10^-3 mm²/s, respectively. No significant differences were observed between the means of the IVIM parameters (D, f, D^*) calculated by the two methods (t = 1.113, -0.259, -1.660; P = 0.272, 0.797, 0.104, respectively), and the relative ICC were 0.863, 0.469, 0.663, respectively. The intra-observer 95% limits of consistency of IVIM parameters were (-0.012-0.038) × 10^-3 mm²/s, (-0.003-0.007), (-0.923-1.166) × 10^-3 mm²/s with ROI outline tumor, respectively; (-0.024-0.044) × 10^-3 mm²/s, (-0.005-0.015), (-1.670-4.195) × 10^-3 mm²/s with center ROI, respectively. The inter-observer 95% limits of consistency of perfusion parameters were (-0.047-0.009) × 10^-3 mm²/s, (-0.015-0.009), (-7.206-3.190) × 10^-3 mm²/s with ROI outlined tumor, respectively; (-0.068-0.048) × 10^-3 mm²/s, (-0.005-0.041), (-17.657-0.779) × 10^-3 mm²/s with center ROI, respectively. Conclusions: There was no statistically significant difference between the outlined ROI and tumor center ROI analysis of rectal cancers' IVIM parameters. The tumor analysis by outlined ROI protocol appropriately improves intra- and inter-observer consistency and can provide more reproducible and stable results. (authors)