Skip to main content
. Author manuscript; available in PMC: 2020 May 27.
Published in final edited form as: Eur Radiol. 2019 Apr 23;29(7):3338–3347. doi: 10.1007/s00330-019-06205-9

Table 3.

Overall accuracy and runtimes for model classification and classification by two radiologists

Accuracy of lesion classification (mean ± SD %) Accuracy of derived LI-RADS classification (mean ± SD %) Runtime (mean ± SD)
Average of 20 iterations
 Model training set 98.7 ± 1.0 99.1 ± 0.7 29 min ± 4
 Model test set 91.9 ± 2.9 94.3 ± 2.9 5.6 ms ± 4.6
Reader study (n = 60)
 Model 90.0 91.7 1.0 ms ± 0.4
 Radiologist 1 80.0 88.3 14 ± 10 s
 Radiologist 2 85.0 88.3 17 ± 24 s