Table 3.

Performance of our model in total kidney volume estimation in magnetic resonance images from dataset C, grouped by scanner manufacturer and model, as compared with reference manual tracing

Manufacturer and Model	No. of Patients	No. of Scans	Correlationa	Total Kidney Volume Difference,b ml	Total Kidney Volume Difference, %	P Valuec
Siemens
Aera	43	109	0.991	−67±185	−1.9±9.9	0.16
Avanto	41	113	0.992	−47±160	−1.9±7.8	0.07
SymphonyTim	29	72	0.991	−52±200	−1.3±8.7	0.58
Avanto fit	12	23	0.995	−29±306	1.3±5.8	0.37
Amira	8	25	0.990	2±81	0.5±5.0	0.84
Other models	43	92	0.994	−63±167	−1.9±7.1	0.01
Philips
Ingenia	26	68	0.998	−41±105	1.4±5.8	0.37
Intera	31	80	0.995	−116±550	−2.3±6.3	0.07
Achieva	24	66	>0.999	−56±134	−1.1±3.9	0.05
Ingenia S	14	40	0.999	−39±75	2.0±4.6	0.17
Achieva dStream	7	20	>0.999	−10±27	0.3±3.1	0.69
Panorama HFO	8	20	0.997	−52±281	2.5±11.1	>0.99
GE
Multiple models	29	73	0.997	−31±108	−0.9±5.8	0.001
TOSHIBA
Multiple models	5	12	0.999	149±378	−0.1±11.0	0.81

High correlations and consistently small differences from reference volumetry can be noticed for the automated method across all of the models.

^aPearson correlation coefficient.

^bTotal kidney volume difference was calculated between automated and manual methods.

^cCalculated with the Wilcoxon signed-rank test.