[Retracted] Models of Artificial Intelligence-Assisted Diagnosis of Lung Cancer Pathology Based on Deep Learning Algorithms

Table 5

Comparison of benign and malignant judgments between the improved 3D U-net, the original 3D U-net, and the physician.

Object	Sensitivity	Specificity	Positive fuel-like ratio	Negative fuel-like ratio
Improved 3D U-net	95.51%97.22–95.46%	34.46%27.86–46.85%	1.211.05–1.39	0.200.10–0.35
3D U-net	91.83%82.70–92.37%	58.69%47.33–66.95%	2.131.47–2.48	0.220.10–0.35
Physician	85.1%76.43–88.67%	70.15%60.22–75.68%	2.812.16–3.64	0.230.10–0.35