Table 2.
Comparison of the oral pathologists' diagnoses with and without deep-learning-assisted diagnoses.
| Evaluator | Normal [95% CI] | SCC [95% CI] | Others [95% CI] | Macro [95% CI] | Micro [95% CI] |
|---|---|---|---|---|---|
| Oral pathologist #1 | |||||
| With assistive diagnosis | 0.97 [0.968–0.979] | 0.99 [0.984–0.989] | 0.94 [0.929–0.942] | 0.97 [0.960–0.966] | 0.97 [0.958–0.965] |
| w/o assistive diagnosis | 0.75 [0.732–0.759] | 0.88 [0.871–0.882] | 0.79 [0.784–0.804] | 0.80 [0.795–0.810] | 0.79 [0.776–0.791] |
| Oral pathologist #2 | |||||
| With assistive diagnosis | 0.98 [0.978–0.986] | 0.94 [0.936–0.947] | 0.88 [0.870–0.885] | 0.93 [0.929–0.939] | 0.93 [0.927–0.937] |
| w/o assistive diagnosis | 0.91 [0.902–0.922] | 0.85 [0.847–0.863] | 0.82 [0.811–0.828] | 0.86 [0.854–0.867] | 0.83 [0.819–0.833] |
| Oral pathologist #3 | |||||
| With assistive diagnosis | 0.73 [0.714–0.741] | 0.91 [0.896–0.909] | 0.83 [0.823–0.839] | 0.82 [0.806–0.818] | 0.88 [0.879–0.889] |
| w/o assistive diagnosis | 0.56 [0.549–0.572] | 0.82 [0.816–0.832] | 0.71 [0.712–0.732] | 0.70 [0.691–0.704] | 0.81 [0.798–0.812] |
| Oral pathologist #4 | |||||
| With assistive diagnosis | 1.00 [0.998–0.999] | 0.98 [0.979–0.984] | 0.96 [0.950–0.959] | 0.98 [0.976–0.980] | 0.98 [0.976–0.982] |
| w/o assistive diagnosis | 0.97 [0.968–0.977] | 0.97 [0.965–0.972] | 0.90 [0.893–0.905] | 0.95 [0.942–0.950] | 0.95 [0.946–0.955] |
| Oral pathologist #5 | |||||
| With assistive diagnosis | 0.96 [0.948–0.964] | 0.95 [0.945–0.953] | 0.92 [0.917–0.928] | 0.94 [0.941–0.949] | 0.96 [0.952–0.959] |
| w/o assistive diagnosis | 0.92 [0.902–0.922] | 0.89 [0.884–0.900] | 0.86 [0.845–0.860] | 0.89 [0.884–0.897] | 0.91 [0.900–0.913] |
| Oral pathologist #6 | |||||
| With assistive diagnosis | 1.00 [0.996–0.997] | 0.98 [0.979–0.968] | 0.97 [0.963–0.968] | 0.98 [0.982–0.985] | 0.98 [0.980–0.983] |
| w/o assistive diagnosis | 0.94 [0.928–0.943] | 0.85 [0.837–0.852] | 0.75 [0.749–0.768] | 0.85 [0.843–0.854] | 0.85 [0.836–0.849] |
95% CI 95% confidence interval.