Skip to main content
. 2022 Jul 13;8(4):1804–1819. doi: 10.3390/tomography8040152

Table 3.

(A) Model accuracy on external validation (n = 30), median (interquartile range). (B) Model accuracy on prospective validation (n = 30), median (IQR).

(A)
External Test Set Right Kidney Left Kidney Liver Spleen
Model volume (mL)
Corrected model volume (mL)
617 (327–1009)
608 (316–1041)
582 (416–1289)
582 (365–1285)
1706 (1292–2087)
1684 (1287–2076)
220 (145–274)
222 (157–280)
DSC 0.96 0.98 0.97 0.96
Concordance Coefficient >0.99 >0.99 0.98 0.99
RMS error (mL) 42 39 258 17
Average % error 7% 3% 3% 1%
Number with zero error 5 (17%) 6 (20%) 1 (3%) 7 (23%)
(B)
Prospective Test Set Right Kidney Left Kidney Liver Spleen
Model volume (mL)
Corrected model volume (mL)
625 (370–1000)
650 (394–998)
729 (481–1039)
768 (485–1043)
1711 (1489–2065)
1727 (1495–2051)
244 (177–315)
241 (175–318)
DSC 0.96 0.96 0.96 0.95
Concordance Coefficient >0.99 >0.99 >0.99 0.98
RMS error (mL) 112 65 112 37
Average % error 6% 5% 5% 1%
# with zero error 2 (7%) 3 (20%) 2 (7%) 2 (7%)