Table 2.
Accuracy of machine learning models in identifying the Raman spectra of gastric cancer cells.
| Dataset | SVM | KNN | LDA | XGBoost | Decision Tree | |
|---|---|---|---|---|---|---|
| AGS | Full Spectrum | 100% | 99.36% | 100% | 92.95% | 100% |
| Fingerprint Region | 100% | 100% | 100% | 94.23% | 98.72% | |
| HW Region | 100% | 100% | 100% | 100% | 99.36% | |
| Background | 96.15% | 94.87% | 96.79% | 90.38% | 87.18% | |
| All Data | 97.44% | 96.79% | 99.36% | 92.95% | 88.46% | |
| BGC-823 | Full Spectrum | 100% | 100% | 100% | 94.84% | 100% |
| Fingerprint Region | 100% | 97.5% | 100% | 89.03% | 92.90% | |
| HW Region | 100% | 99.35% | 100% | 99.35% | 100% | |
| Background | 96.13% | 94.19% | 89.03% | 91.61% | 91.61% | |
| All Data | 96.77% | 97.5% | 100% | 94.84% | 100% | |
| HGC-27 | Full Spectrum | 100% | 100% | 100% | 95.14% | 94.44% |
| Fingerprint Region | 100% | 100% | 100% | 95.83% | 100% | |
| HW Region | 100% | 100% | 100% | 93.06% | 100% | |
| Background | 98.61% | 95.83% | 96.53% | 95.14% | 93.75% | |
| All Data | 100% | 98.61% | 100% | 95.14% | 97.22% | |
| MKN-45 | Full Spectrum | 97.14% | 95.39% | 99.34% | 87.5% | 94.08% |
| Fingerprint Region | 99.34% | 99.34% | 100% | 96.71% | 88.16% | |
| HW Region | 92.76% | 98.03% | 100% | 95.39% | 95.39% | |
| Background | 96.71% | 87.5% | 98.68% | 84.87% | 91.48% | |
| All Data | 98.03% | 90.13% | 99.34% | 87.5% | 91.48% | |
| MKN-74 | Full Spectrum | 100% | 100% | 100% | 93.24% | 100% |
| Fingerprint Region | 100% | 99.32% | 100% | 100% | 100% | |
| HW Region | 100% | 100% | 100% | 95.95% | 97.97% | |
| Background | 95.95% | 95.27% | 99.32% | 92.27% | 95.27% | |
| All Data | 96.62% | 95.95% | 100% | 93.34% | 94.59% | |
| SNU-16 | Full Spectrum | 100% | 100% | 100% | 91.88% | 96.25% |
| Fingerprint Region | 100% | 99.38% | 100% | 86.88% | 91.25% | |
| HW Region | 100% | 100% | 100% | 100% | 98.75% | |
| Background | 99.38% | 98.75% | 97.5% | 96.25% | 94.38% | |
| All Data | 98.75% | 98.75% | 100% | 91.88% | 92.5% |