Skip to main content
. Author manuscript; available in PMC: 2022 Jan 1.
Published in final edited form as: Med Image Anal. 2020 Sep 25;67:101814. doi: 10.1016/j.media.2020.101814

Table 2.

Comparison of the performance of the competing algorithms in the form of AUC and AP scores.

Method AUC AP
CNN (Y,S) 0.748 ± 0.035 0.498 ± 0.037
NoisyAND (Y,S) 0.761 ± 0.027 0.538 ± 0.037
AttentionMIL (Y,S) 0.743 ± 0.055 0.486 ± 0.095

NoisyAND (U,Y,S) 0.845 ± 0.016 0.708 ± 0.041
AttentionMIL (U,Y,S) 0.823 ± 0.021 0.643 ± 0.048
AveragePooling (U,Y,S) 0.850 ± 0.025 0.693 ± 0.037

Proposed (U,Y) 0.858 ± 0.017 0.719 ± 0.029
Proposed (U,S) 0.852 ± 0.024 0.713 ± 0.040
Proposed (U,V) 0.835 ± 0.024 0.693 ± 0.049
Proposed (U,Y,V) 0.858 ± 0.014 0.721 ± 0.035
Proposed (U,S,V) 0.857 ± 0.018 0.733 ± 0.048
Proposed2Heads (U,Y,S) 0.860 ± 0.019 0.711 ± 0.046
Proposed (U,Y,S) 0.870 ± 0.017 0.743 ± 0.037
Proposed (U,Y,S,V) 0.860 ± 0.024 0.730 ± 0.047