Table 3: Comparison and correlation of the ConvNet UHCMC/CWRU and ConvNet HUP classifiers in terms of Dice, PPV, NPV, TPR, TNR, FPR and FNR.

From: Accurate and reproducible invasive breast cancer detection in whole-slide images: A Deep Learning approach for quantifying tumor extent

 DicePPVNPVTPRTNRFPRFNR
TCGA
ConvNetHUP0.7494 ± 0.20710.7071 ± 0.22540.9658 ± 0.05140.8600 ± 0.17050.9188 ± 0.08050.0812 ± 0.08050.1400 ± 0.1705
ConvNetUHCMC/CWRU0.7068 ± 0.20610.6464 ± 0.21880.9629 ± 0.05840.8676 ± 0.17060.8880 ± 0.08240.1120 ± 0.08240.1324 ± 0.1706
r0.87330.92580.81090.63450.80550.80550.6345
NC
ConvNetHUPN/AN/A1 ± 0N/A0.9716 ± 0.06930.0284 ± 0.0693N/A
ConvNetUHCMC/CWRUN/AN/A1 ± 0N/A0.9546 ± 0.08160.0454 ± 0.0816N/A
rN/AN/AN/AN/A0.68760.6876N/A
  1. Note that for the normal cases considered, not all the performance measures are shown because the NC data cohort did not have cancer annotations.