A systematic review of radiomics in pancreatitis: applying the evidence level rating tool for promoting clinical transferability

Insights into Imaging

Table 4 Diagnostic performance of meta-analyzed clinical questions

Clinical question	AIP versus PC by CT	MFP versus PC by MRI
Number of studies	6	4
Number of available datasets	5/8	5/6
Events/sample size	191/421	101/320
Pooled analysis
DOR (95% CI)	189.63 (79.65–451.48)	135.70 (36.17–509.13)
p value for DOR	< 0.001	< 0.001
Sensitivity (95% CI)	0.90 (0.84–0.94)	0.90 (0.81–0.95)
Specificity (95% CI)	0.95 (0.92–0.97)	0.94 (0.86–0.98)
PLR (95% CI)	19.01 (10.51–34.40)	15.00 (5.94–37.92)
NLR (95% CI)	0.10 (0.06–0.17)	0.11 (0.06–0.56)
AUC (95% CI)	0.97 (0.95–0.98)	0.95 (0.93–0.96)
Heterogeneity
Higgins I² test (%)	83.26%	97.28%
Cochran’s Q test (p value)	< 0.01	< 0.01
Publication bias
Egger’s test (p value)	0.060	0.050
Begg’s test (p value)	0.221	0.221
Deeks test (p value)	0.226	0.538
Trim and fill method
Number of missing datasets	2	2
Adjusted DOR (95%CI)	135.11 (64.40–283.74)	53.89 (15.95–182.00)
Level of evidence	Weak	Weak

AIP autoimmune pancreatitis, AUC area under curve, CI confidential interval, DOR diagnostic odds ratio, MFP mass-forming pancreatitis, NLR negative likelihood ratio, n/a not applicable, PC pancreatic cancer, PLR positive likelihood ratio