Table 1.
Dataset | Group | AUC | Accuracy | Sensitivity | Specificity | PPV | NPV | χ2 | p value* | |
---|---|---|---|---|---|---|---|---|---|---|
SLAP-Net | Dataset 1** | Class 0 | 0.98 | 0.96 | 0.94 | 1.00 | 1.00 | 0.91 | / | / |
Class 1 | 0.98 | 0.96 | 1.00 | 0.94 | 0.91 | 1.00 | / | / | ||
Dataset 2** | Class 0 | 0.92 | 0.85 | 0.90 | 0.76 | 0.86 | 0.84 | / | / | |
Class 1 | 0.92 | 0.85 | 0.76 | 0.90 | 0.84 | 0.86 | / | / | ||
Radiologist 1 (15 years of experience) | Dataset 1 | Class 0 | / | 0.85 | 0.91 | 0.76 | 0.86 | 0.84 | 3.68 | 0.055 |
Class 1 | / | 0.85 | 0.76 | 0.91 | 0.84 | 0.86 | ||||
Dataset 2 | Class 0 | / | 0.85 | 0.86 | 0.84 | 0.91 | 0.77 | 0.53 | 0.468 | |
Class 1 | / | 0.85 | 0.84 | 0.86 | 0.77 | 0.91 | ||||
Radiologist 2 (10 years of experience) | Dataset 1 | Class 0 | / | 0.83 | 0.81 | 0.85 | 0.90 | 0.74 | 5.04 | 0.025 |
Class 1 | / | 0.83 | 0.85 | 0.81 | 0.74 | 0.90 | ||||
Dataset 2 | Class 0 | / | 0.86 | 0.89 | 0.82 | 0.90 | 0.80 | 1.13 | 0.289 | |
Class 1 | / | 0.86 | 0.82 | 0.89 | 0.80 | 0.90 | ||||
Radiologist 3 (7 years of experience) | Dataset 1 | Class 0 | / | 0.81 | 0.78 | 0.85 | 0.89 | 0.71 | 6.50 | 0.011 |
Class 1 | / | 0.81 | 0.85 | 0.78 | 0.71 | 0.89 | ||||
Dataset 2 | Class 0 | / | 0.81 | 0.86 | 0.73 | 0.85 | 0.74 | 0.47 | 0.495 | |
Class 1 | / | 0.81 | 0.73 | 0.86 | 0.74 | 0.85 |
AUC area under the receiver operating characteristic curve, PPV positive predictive value, NPV negative predictive value, χ2 the chi-square value calculated by McNemar’s test
*The results of comparing different radiologists with SLAP-Net on dataset 1 and dataset 2; McNemar’s test was used for the statistical analysis
**Dataset 1 contains 52 test patients, and dataset 2 is an independent test set that contains 122 patients