Table 2.

Results of three models: AUC value, accuracy, sensitivity, and specificity

Methods	Medical image	Performance
		AUC (95% CI)	Accuracy (95% CI)	Sensitivity (95% CI)	Specificity (95% CI)
RAD_model	CT	82.7% ± 7.3% (76.2–89.1%)	73.0% ± 5.6% (68.1–77.9%)	76.2% ± 7.6% (69.5–82.9%)	70.4% ± 10.7% (61.0–79.8%)
	PET	80.8% ± 6.1% (75.4–86.1%)	73.4% ± 7.1% (67.2–79.6%)	66.4% ± 9.5% (58.1–74.8%)	81.2% ± 13.2% (69.7–92.8%)
	PET/CT	89.5% ± 6.1% (84.2–94.8%)	80.0% ± 7.1% (73.8–86.3%)	75.2% ± 9.3% (67.1–83.4%)	85.5% ± 11.6% (75.4–95.7%)
DL_model	CT	90.4% ± 1.0% (89.6–91.3%)	81.0% ± 1.4% (79.8–82.3%)	79.2% ± 5.3% (74.6–83.8%)	83.2% ± 4.5% (79.2–87.2%)
	PET	86.7% ± 4.1% (83.2–90.3%)	79.7% ± 5.0% (75.3–84.0%)	75.6% ± 10.4% (66.5–84.7%)	84.0% ± 5.3% (79.4–88.6%)
	PET/CT	93.6% ± 1.5% (92.3–94.9%)	85.8% ± 1.7% (84.3–87.2%)	84.4% ± 8.1% (77.3–91.5%)	87.9% ± 7.3% (81.5–94.3%)
MF_model	CT	92.6% ± 1.6% (91.2–94.1%)	84.3% ± 2.5% (82.1–86.4%)	85.4% ± 9.3% (77.3–93.5%)	83.2% ± 8.3% (75.9–90.5%)
	PET	92.7% ± 2.5% (90.5–94.9%)	82.8% ± 6.0% (77.5–88.1%)	73.8% ± 15.0% (60.6–86.9%)	91.6% ± 4.8% (87.4–95.8%)
	PET/CT	96.4% ± 1.1% (95.4–97.3%)	90.1% ± 1.6% (88.7–91.5%)	87.5% ± 3.6% (84.3–90.6%)	93.0% ± 3.0% (90.3–95.6%)