TABLE 4. Quantitative Results Associated With Training, Testing Errors, of 3DDCNN and Cloud-Based 3DDCNN CAD System on Different Datasets.
| Dataset | 3DDCNN CAD Mode | Training Error | Testing Error | Sensitivity | FP/scan |
|---|---|---|---|---|---|
| LIDC-IDRI | Stand-Alone | 0.00448 | 0.00661 | 0.952 | 2.4 |
| [22] | Cloud-Based | 0.00391 | 0.00398 | 0.974 | 2.1 |
| ANODE09 | Stand-Alone | 0.00542 | 0.00478 | 0.950 | 2.9 |
| [21] | Cloud-Based | 0.00419 | 0.00354 | 0.976 | 2.3 |
| LUNA16 | Stand-Alone | 0.00391 | 0.00328 | 0.961 | 2.2 |
| [20] | Cloud-Based | 0.00368 | 0.00274 | 0.988 | 1.97 |
| SPH6 | Stand-Alone | 0.00364 | 0.00288 | 0.964 | 2.1 |
| Dataset | Cloud-Based | 0.00294 | 0.00224 | 0.987 | 1.97 |
”SPH6” refers to Clinical Dataset from Shanghai Sixth People’s Hospital