Table 2.

Performance assessment.

Algorithm	AUC from ROC analysis	Human-engineered features	Deep CNN features	Interpretable	Transfer learning
Transfer AMIL	0.94 ± 0.04		X	X	X
Noisy 2D Classifier	0.85 ± 0.06		X
Fully 3D Classifier	0.58 ± 0.16		X
AMIL	0.69 ± 0.05		X	X
Mean pooling	0.90 ± 0.02		X		X
Max pooling	0.88 ± 0.02		X		X
Cheplygina20	0.78 ± 0.04	X		X
Orting21	0.88 ± –	X
Tennakoon22	0.95 ± –		X