Table 2 Heterogeneity analysis results included in the study.
Subgroup | No. of studies | Sensitivity (95% CI) | Specificity (95% CI) | AUC (95% CI) | DLR+ (95% CI) | DLR− (95% CI) | DOR (95% CI) |
|---|---|---|---|---|---|---|---|
Country | |||||||
China | 16 | 0.80 (0.73–0.86) | 0.85 (0.81–0.89) | 0.90 (0.87–0.92) | 5.36 (4.0–7.05) | 0.23 (0.16–0.32) | 23.22 (14.04–38.40) |
Other | 5 | 0.85 (0.53–0.97) | 0.85 (0.71–0.93) | 0.91 (0.88–0.93) | 5.81 (2.36–14.32) | 0.17 (0.04–0.77) | 33.86 (3.37–340.56) |
Modeling method | |||||||
Deep learning | 4 | 0.80 (0.56–0.92) | 0.79 (0.71–0.85) | 0.924 (0.872–0.984) | 3.82 (2.12–6.09) | 0.25 (0.09–0.59) | 15.17 (3.67–62.55) |
Radiomics | 12 | 0.83 (0.76–0.88) | 0.83 (0.78–0.87) | 0.90 (0.87–0.92) | 4.81 (3.76–6.16) | 0.21 (0.15–0.29) | 23.04 (14.33–37.04) |
Machine learning | 5 | 0.80 (0.51–0.94) | 0.91 (0.82–0.96) | 0.94 (0.91–0.95) | 8.78 (3.56–21.70) | 0.22 (0.07–0.69) | 40.65 (5.79–285.42) |
Sample size | |||||||
> 100 | 9 | 0.82 (0.67–0.91) | 0.86 (0.79–0.91) | 0.91 (0.88–0.93) | 6.06 (3.62–10.14) | 0.21 (0.10–0.42) | 29.25 (9.70–88.19) |
< 100 | 12 | 0.81 (0.73–0.88) | 0.83 (0.78–0.88) | 0.89 (0.86–0.92) | 4.92 (3.60–6.71) | 0.22 (0.15–0.33) | 22.05 (11.76–41.33) |
MRI sequence | |||||||
Multi-parametric | 15 | 0.81 (0.72–0.88) | 0.85 (0.80–0.89) | 0.90 (0.87–0.93) | 5.49 (3.91–7.73) | 0.22 (0.14–0.34) | 25.23 (12.83–49.64) |
Enhanced T1WI | 6 | 0.85 (0.68–0.94) | 0.86 (0.77–0.92) | 0.92 (0.89–0.94) | 6.03 (3.24–11.23) | 0.17 (0.07–0.43) | 35.71 (7.89–156.82) |
Combine clinical factors | |||||||
Yes | 4 | 0.83 (0.76–0.88) | 0.88 (0.77–0.94) | 0.84 (0.81–0.87) | 6.79 (3.51–13.12) | 0.19 (0.13–0.29) | 34.94 (14.72–82.90) |
No | 17 | 0.82 (0.72–0.89) | 0.85 (0.80–0.89) | 0.90 (0.87–0.93) | 5.45 (3.90–7.62) | 0.21 (0.13–0.34) | 26.16 (12.48–54.83) |
