Research Article
Comparative Study on Automated Cell Nuclei Segmentation Methods for Cytology Pleural Effusion Images
Table 1
Quantitative experimental results.
| Methods/evaluation | Pre | Re | Fm | JI | DSC (%) |
| Otsu’s method | 0.91 | 0.89 | 0.90 | 0.89 | 94 | Isodata | 0.86 | 0.84 | 0.85 | 0.84 | 91 | Maximum entropy | 0.84 | 0.94 | 0.88 | 0.94 | 91 | Cross entropy | 0.94 | 0.82 | 0.87 | 0.82 | 90 | Minimum error | 0.85 | 0.82 | 0.83 | 0.82 | 89 | Fuzzy entropy | 0.68 | 0.67 | 0.68 | 0.67 | 80 | Adaptive thresholding | 0.96 | 0.50 | 0.66 | 0.50 | 67 | k-means | 0.90 | 0.89 | 0.89 | 0.89 | 94 | Fuzzy c-means | 0.94 | 0.77 | 0.85 | 0.77 | 77 | Mean shift | 0.93 | 0.91 | 0.92 | 0.91 | 95 | Chan–Vese method | 0.89 | 0.87 | 0.88 | 0.87 | 94 | Graph-based min cut | 0.87 | 0.95 | 0.91 | 0.87 | 93 |
|
|