Meeting Abstract
Enhancing Automatic Classification of Hepatocellular Carcinoma Images through Image Masking, Tissue Changes, and Trabecular Features
Table 1
Experiment results.
| Training data | Test data | Set of features | Sensitivity | Specificity |
| Biopsy sample | Biopsy sample | 78 nuclei (unmasked) | 86.36% | 88.29% | Surgery sample | Surgery sample | 78 nuclei (unmasked) | 88.21% | 87.99% | Combination of biopsy and surgery sample | Biopsy sample | 78 nuclei (unmasked) | 84.73% | 91.87% | 78 nuclei (masked) | 85.27% | 90.67% | 72 nuclei (masked) + 21 new features | 88.18% | 91.87% | Surgery sample | 78 nuclei (unmasked) | 88.95% | 87.62% | 78 nuclei (masked) | 89.50% | 87.62% | 72 nuclei (masked) + 21 new features | 91.34% | 89.68% |
|
|