Research Article
Convolutional Neural Networks for Recognition of Lymphoblast Cell Images
Table 6
Sensitivity of ConVNet and SVM-GA to identify ALL subtypes over ten test sets.
| Test set | ConVNet | SVM-GA | Lymphocyte | Pre-T | Pre-B | Lymphocyte | Pre-T | Pre-B |
| 1 | 100.00 | 71.11 | 82.22 | 100.00 | 77.78 | 73.33 | 2 | 100.00 | 66.67 | 80.00 | 93.55 | 80.00 | 73.33 | 3 | 100.00 | 75.56 | 84.44 | 93.55 | 66.67 | 84.44 | 4 | 100.00 | 64.44 | 82.22 | 100.00 | 77.78 | 73.33 | 5 | 100.00 | 62.22 | 80.00 | 90.32 | 60.00 | 91.11 | 6 | 100.00 | 57.78 | 84.44 | 83.87 | 80.00 | 82.22 | 7 | 100.00 | 80.00 | 75.56 | 96.77 | 75.56 | 75.56 | 8 | 100.00 | 82.22 | 80.00 | 90.32 | 77.78 | 80.00 | 9 | 96.77 | 71.43 | 80.00 | 96.77 | 84.44 | 82.22 | 10 | 100.00 | 57.78 | 86.67 | 100.00 | 68.89 | 84.44 | Average | 99.68 ± 1.02 | 68.92 ± 8.65 | 81.56 ± 3.15 | 94.52 ± 5.28 | 74.89 ± 7.41 | 80.00 ± 6.02 |
|
|