TABLE 6.
AI in the Diagnosis of Gastric Cancer and CAG
| Performance | |||||||
|---|---|---|---|---|---|---|---|
| References | AI Model | Study type | Aim | Diagnostic Tool | Accuracy | Sensitivity | Specificity |
| Li et al84 | DL | Retrospective | Diagnosis of GC | ME-NBI | 90.9% | 91.1% | 90.6% |
| Hirasawa et al82 | DL | Retrospective | Diagnosis of GC | WLE, NBI, IC | — | 92.2%-98.6% | — |
| Wu et al85 | DL | Retrospective | Diagnosis of GC | WLE, NBI, BLI | 92.5% | 94.0% | 91.0% |
| Ueyama et al86 | DL | Retrospective | Diagnosis of GC | ME-NBI | 98.7% | 98.0% | 100% |
| Zhang et al87 | DL | Retrospective | Detection of CAG | WLE | 94.0% | 95.0% | 94.0% |
| Lee et al88 | DL | Retrospective | Differential diagnosis GC vs. gastric ulcer | WLE | 77.1%-90% | — | — |
| Horiuchi et al89 | DL | Retrospective | Differential diagnosis GC vs. gastritis | ME-NBI | 85.3% | 95.4% | 71.0% |
| Zhu et al90 | DL | Retrospective | Characterization of GC invasion depth | WLE | 89.1% | 76.5% | 95.5% |
| Nagao et al91 | DL | Retrospective | Characterization of GC invasion depth | WLE, NBI, IC | 94.5% | 84.4 | 99.4% |
AI indicates artificial intelligence; BLI, blue-laser imaging; CAG, chronic atrophic gastritis; DL, deep learning; GC, gastric cancer; IC, indigo-carmine dye contrast; ME, magnified endoscopy; NBI, narrow-bad imaging; WLE, white-light endoscopy.