Table 1.
Artificial intelligence applications on lung computed tomography scans and chest X-rays for COVID-19 diagnosis
| Patient population | Task | Type/number of images | Methods | Results |
|---|---|---|---|---|
| 3322 patients from six medical centers in Wuhan, Shandong, Beijing, and Shenzhen, China [5] | Classification for detecting COVID-19-positive cases | 4356 3D chest CT scans | Deep learning model, COVID-19 detection neural network (COVNet) | AUC 0.96 (P- value < 0.001) |
| 905 patients from 18 medical centers in 13 provinces in China [6▪▪] | Rapidly diagnose patients who are COVID-19 positive by integrating chest CT findings with clinical symptoms, exposure history, and laboratory testing | Chest CT scans | Three AI algorithms: CNN for only CT scans; ML: SVM, Random forest for clinical data; joint CNN model combining CT scans and clinical data | AUC 0.92 AUC (95% CI 0.887–0.948) joint CNN model |
| 125 COVID-19-positive cases from open source chest X-ray dataset. Negative cases from another dataset (32717 unique patients with disease labels) [7] | Binary classification (COVID vs. no-findings) and multiclass classification (COVID vs. no-findings vs. pneumonia) | 1000 Chest X-ray images | Deep learning. DarkNet architecture | AUC 0.98 for (COVID vs. no-findings) and 0.87 for multiclass (COVID vs. no-findings vs. pneumonia) |
| 454 patients in the Netherlands (223 positive cases and 231 negative cases) [8] | Classification for detecting COVID-19 cases | 24678 Chest X-Ray images | AI system (CAD4COVID-Xray) | AUC 0.81 |
| 1186 patients: 521 COVID-19 positive cases from Rhode Island Hospital and 9 Hospitals in Hunan Province, China 665 with non-COVID-19 pneumonia from Rhode Island Hospital, University of Pennsylvania and Xiangya Hospital [9] | Classification of COVID-19 cases | 1186 CT scans 132583 CT slices |
EfficientNet B4 deep neural network architecture after lung segmentation, followed by two-layer fully connected neural network to pool slices together | AUC 0.96 (95% CI 0.9–0.98) |
AUC, Area Under the Curve; CNN, Convolutional Neural Network; CT, computed tomography; ML, Machine Learning; SVM, Support Vector Machine.