TABLE 2. The Accuracy, Sensitivity, Specificity of Case-Based COVID-19 Diagnosis of Our Method Compared to Mask-R-CNN -SVM Method.
Method | Source | Image Sets | Accuracy | Sensitivity | Specificity |
---|---|---|---|---|---|
Mask-R-CNN-SVM | Test | vs Lung Nodules | 0.744 | 0.649 | 0.829 |
vs Other Pneumonia | 0.698 | 0.649 | 0.769 | ||
External Validation | vs Other Pneumonia | 0.599 | 0.575 | 0.667 | |
vs Without Known Lung Diseases | 0.626 | 0.646 | 0.719 | ||
NCIP-Net | Test | vs Lung Nodules | 0.861 | 0.757 | 0.952 |
Vs Other Pneumonia | 0.757 | 0.757 | 0.815 | ||
External Validation | vs Other Pneumonia | 0.744 | 0.713 | 0.861 | |
vs Without Known Lung Diseases | 0.829 | 0.713 | 0.959 |
note that we used the case-based probability of 0.5 as the thresholding value to generate all of the results in this table.