Skip to main content

Table 4 Comparison of direct/model-aided diagnostic performance of the gynecologists in the five-category classification task

From: Deep learning model for classifying endometrial lesions

Category Sensitivity (%) Specificity (%) Precision (%) F1-score (%) AUC Accuracy (%)
Gynecologist 4
 EH 59.1 94.2 68.4 63.4 0.766 64.0
 AH 58.6 93.8 73.9 65.4 0.762
 EC 56.5 89.2 54.2 55.3 0.729
 EP 78.3 85.3 54.5 64.3 0.818
 SM 67.9 92.8 73.1 70.4 0.803
Model-aided gynecologist 4
 EH 82.1 93.8 79.3 80.7 0.880 78.4
 AH 47.6 98.1 83.3 60.6 0.728
 EC 77.8 94.9 80.8 79.2 0.863
 EP 92.6 90.8 73.5 82.0 0.917
 SM 86.4 95.1 79.2 82.6 0.908
Gynecologist 5
 EH 68.0 82.5 45.5 54.5 0.754 62.4
 AH 55.2 93.8 72.7 62.7 0.745
 EC 52.2 90.2 54.5 53.3 0.712
 EP 73.9 88.2 58.6 65.4 0.811
 SM 64.3 99.0 94.7% 76.6 0.816
Model-aided gynecologist 5
 EH 82.1 84.5 60.5 69.7 0.833 72.8
 AH 42.9 92.3 52.9 47.4 0.676
 EC 74.1 96.9 87.0 80.0 0.855
 EP 81.5 94.9 81.5 81.5 0.882
 SM 77.3 97.1 85.0 81.0 0.872
Gynecologist 6
 EH 59.1 96.1 76.5 66.7 0.776 69.2
 AH 58.6 85.4 54.8 56.7 0.720
 EC 52.2 92.2 60.0 55.8 0.722
 EP 91.3 93.1 75.0 82.4 0.922
 SM 78.6 92.8 75.9 77.2 0.857
Model-aided gynecologist 6
 EH 75.0 96.9 87.5 80.8 0.860 77.6
 AH 57.1 94.2 66.7 61.5 0.757
 EC 77.8 93.9 77.8 77.8 0.858
 EP 88.9 89.8 70.6 78.7 0.893
 SM 86.4 97.1 86.4 86.4 0.917
  1. AH atypical hyperplasia, AUC area under the receiver operating characteristic (ROC) curve, EC endometrial cancer, EH endometrial hyperplasia without atypia, EP endometrial polyp, SM submucous myoma