Research Article
A Computer-Aided Diagnosis System Using Deep Learning for Multiclass Skin Lesion Classification
Table 4
Classification results using proposed feature selection algorithm on augmented HAM10000 dataset.
| Classifier (ESD) | Recall rate (%) | Precision rate (%) | FNR (%) | AUC | Accuracy (%) | Time (sec) | F1-score (%) |
| 500 | 78.42 | 79.28 | 21.57 | 0.954 | 78.3 | 102.4 | 78.85 | 1000 | 64.85 | 66.71 | 35.14 | 0.902 | 65.1 | 132.5 | 65.77 | 1500 | 87.14 | 87.57 | 12.85 | 0.978 | 86.8 | 406.0 | 87.35 | 2000 | 83.57 | 84.42 | 16.42 | 0.97 | 83.3 | 748.4 | 83.99 | 2500 | 89.71 | 90.14 | 10.28 | 0.984 | 89.5 | 1293.5 | 89.92 | 3000 | 91.85 | 92.00 | 8.142 | 0.99 | 91.7 | 1367.6 | 91.92 |
|
|
The bold value represents best ones.
|