Skip to main content
. Author manuscript; available in PMC: 2020 Aug 1.
Published in final edited form as: Magn Reson Med. 2019 Apr 8;82(2):786–795. doi: 10.1002/mrm.27758

Table 2.

Performance metrics for SLN prediction based on the radiomic model trained by R3’s ROIs in the training dataset alongside testing set performance with R1, R2, R3 and CNN segmentations trained by R3’s ROIs.

Training set R3 Testing set
R1 R2 R3 CNN (trained by R3)*
AUC 0.908 0.826 0.841 0.847 0.880
Sensitivity 0.910 0.722 0.778 0.722 0.722
Specificity 0.736 0.778 0.778 0.861 0.889
NPV 0.898 0.848 0.875 0.861 0.865
*

p=0.395, DeLong test between AUCs of R3 and CNN trained by R3.