Table 8. Deep learning model accuracy, specificity and sensitivity for WBC classification (n = 36).
| Author | Year | Method | Accuracy (%) | Sensitivity (%) | Specificity (%) |
|---|---|---|---|---|---|
| Macawile et al. [39] |
2018 | AlexNet | 96.63 | 98.85 | 99.61 |
| Liang et al. [73] | 2018 | CNN + RNN | 91 | - | - |
| Sharma et al. [41] | 2019 | CNN | 97 | 94 | 98 |
| Togacar et al. [42] | 2019 | CNN | 97.78 | - | - |
| Mohamed et al. [74] | 2020 | Pre-trained Deep Learning Models |
97.03 | 71 | 91 |
| Ergen et al. [33] | 2020 | CNN, Feature Selection | 97.95 | 98 | 97.75 |
| Zhao et al. [75] | 2021 | TWO-DCNN | 96 | - | - |
| Cinar et al. [76] | 2021 | Alexnet- GoogleNet-SVM | 99.73, 98.23 |
98.75 | - |
| Wang et al. [77] | 2019 | CNN Architecture SSD and YOLOv3 | 90.09 | - | - |
| Kutlu et al. [14] | 2020 | R-CNN | 97.52 | 88.9 | - |
| Fan et al. [78] | 2019 | ResNet50 | 98 | - | - |
| Hegde et al. [79] | 2019 | Pre-trained AlexNet model | 98.9 | 98.6 | 98.7 |
| Acevedo et al. [80] | 2019 | Pre trained CNN | 96.2 | - | - |
| Qin et al. [81] | 2018 | Deep Residual Learning | 76.84 | - | - |
| Tiwari et al. [82] | 2018 | Double CNN model | 97 | 83 | - |
| Hung et al. [83] | 2017 | AlexNet and Fast R CNN Model | 72 | - | - |
| Naz et al. [84] | 2017 | CNN, faster R CNN | 94.71 | 95.42 | 99.27 |
| Vogado et al. [85] | 2018 | CNN with SVM | 99.20 | 99.2 | - |
| Habibzadeh et al. [86] | 2018 | ResNet and Inception | 99.46 | 99.89 | - |
| Song et al. [87] | 2014 | CNN | 94.5 | 87.26 | - |
| Fatih et al. [88] | 2019 | MRMR feature selection -ELM and CNN | 97.37 | - | - |
| Rehman et al. [89] | 2018 | Deep CNN | 97.78 | - | - |
| Bani-Hani et al. [90] | 2018 | CNN with the optimized genetic method | 91 | 91 | 97 |
| Di Ruberto et al. [91] | 2020 | Pre trained AlexNet | 97.93 | 99.6 | - |
| Loey et al. [92] | 2020 | Pre trained CNN AlexNet | 100 | 100 | 98.2 |
| Ma et al. [93] | 2020 | Generative Adversarial Network and residual neural network | 91.7 | 92 | - |
| Baydilli et al. [94] | 2020 | Capsule Networks | 96.86 | 92.5 | 98.6 |
| Tobias et al. [95] | 2020 | Faster Residual Neural Network | 83.25 | - | - |
| Kassani et al. [96] | 2019 | Hybrid DL based model | 96.17 | 95.17 | 98.58 |
| Baghel et al. [97] | 2022 | CNN | 98.51 | 98.4 | - |
| Shahzad et al. [98] | 2022 | CNN | 98.44 | 99.96 | 99.98 |
| C. Cheuque et al. [99] | 2022 | Multilevel CNN | 98.4 | 98.3 | - |
| Hosseini et al. [100] | 2022 | Convolutional Neural Network | 97 | 94 | 98 |
| Ramya et al. [101] | 2022 | CNN-PSO | 99.2 | 94.56 | 98.78 |
| Khalil et al. [102] | 2022 | CNN | 98 | - | - |
| Sharma et al. [103] | 2022 | DenseNet121 | 98.84 | 98.85 | 99.61 |