Cascaded-Recalibrated Multiple Instance Deep Model for Pathologic-Level Lung Cancer Prediction in CT Images

Table 5

Performance comparison of lung cancer prediction (pmal) in terms of accuracy and AUC.

Method	Accuracy	AUC
Radiologists’ ratings (Mal ⟶ Pmal)	0.7106	0.7621
DARS [11]	0.7501	—
DRS [12]	0.7752	—
CNN nodule(Mal ⟶ Pmal) [8]	0.6538	0.63
CNN-MIL (Mal ⟶ Pmal) [8]	0.7069 ± 0.02	0.66 ± 0.03
Higher-order transfer (Mal ⟶ Pmal) [13]	0.7019 ± 0.05	0.6688 ± 0.05
Higher-order transfer(Tex ⟶ Pmal) [13]	0.7677 ± 0.07	0.7293 ± 0.07
Higher-order transfer(Tex + Diam + Lob ⟶ Pmal) [13]	0.8194 ± 0.02	0.7533 ± 0.05
Recalibrated MIL(Mal ⟶ Pmal) (ours)	0.837 ± 0.035	0.830 ± 0.021
Recalibrated MIL(Sph ⟶ Pmal) (ours)	0.848 ± 0.059	0.831 ± 0.080
Recalibrated MIL(Tex ⟶ Pmal) (ours)	0.846 ± 0.037	0.849 ± 0.027
Cascaded-recalibrated MIL(Tex + Sph ⟶ Pmal) (ours)	0.863 ± 0.051	0.867 ± 0.070
Cascaded-recalibrated MIL(Tex + Sph + Mal ⟶ Pmal) (ours)	0.880 ± 0.032	0.877 ± 0.036

The bold values mean the superiority to others.