Table 3 Variation in performance across scanners by means of multiple linear regression analyses (n = 42; n = 7 per scanner).

Method	GE Signa HDxt 1.5T	GE Signa HDxt 3T	GE Discovery MR750 3T	Philips Ingenuity 3T	Philips Ingenia 3T	Philips Achieva 3T
Cascade	0.09 [−0.09; 0.27]	−0.06 [−0.24; 0.12]	−0.08 [−0.26; 0.10]	0.03 [−0.15; 0.21]	0.003 [−0.18; 0.18]	0.01[−0.17; 0.19]
kNN-TTP	0.01 [−0.08; 0.10]	−0.06 [−0.15; 0.03]	−0.03 [−0.12; 0.07]	0.02 [−0.08; 0.11]	0.03 [−0.06; 0.12]	0.03 [−0.06; 0.12]
Lesion-TOADS	0.12 [−0.08; 0.33]	0.04 [−0.17; 0.24]	−0.05 [−0.26; 0.16]	−0.12 [−0.33; 0.08]	0.10 [−0.11; 0.30]	−0.08 [−0.29; 0.12]
LST-LGA	0.02 [−0.11; 0.14]	−0.04 [−0.17; 0.09]	−0.03 [−0.16; 0.10]	−0.04 [−0.17; 0.09]	0.07 [−0.05; 0.20]	0.02 [−0.10; 0.15]
LST-LPA	0.06 [−0.07; 0.20]	−0.10 [−0.24; 0.03]	−0.09 [−0.23; 0.05]	−0.01 [−0.15; 0.13]	0.11 [−0.03; 0.24]	0.03 [−0.10; 0.17]

Data are represented as unstandardized beta coefficients with 95% confidence intervals. We assessed whether the DSC (as an outcome) depended on scanner (as a categorical variable with each scanner being compared to all other scanners as the reference) using linear regression analysis. A significant relation between a certain scanner and the DSC (family wise error rate corrected p-value of <0.05 using a Bonferroni correction) indicates that the performance (in terms of spatial correspondence with the reference segmentation) was biased for that segmentation method by the use of that scanner (compared to the other scanners). As can be seen in the table, no significant relations were seen for any of the methods.

Quick links

Search