Research Article
Security and Privacy of Cloud- and IoT-Based Medical Image Diagnosis Using Fuzzy Convolutional Neural Network
Table 4
Recent medical applications of CNN-based methods.
| Method | Dataset | Number of images/classes | Application | Accuracy (%) |
| Convolutional neural network [43] | Kaggle dataset | 80000 images, 5 classes | Diabetic retinopathy | 75 | Pretrained convolutional neural network [44] | Ultrasound images | 15000 ultrasound images, 2 classes | Thyroid nodule diagnosis | 83 | Convolutional restricted Boltzmann machine [45] | ILD (interstitial lung diseases) CT scans | 73 CT scans, 5 classes | Lung texture classification and airway detection | 89 | Fuzzy-based pooling in convolutional neural network [46] | MNIST dataset | 60000 grayscale images, 10 classes | Image classification | 94.4 | CIFAR-10 dataset | 60000 RGB images, 10 classes | 27.92 | Tree-based convolutional neural network [19] | NWPU-VHR-10 dataset | More than 60000 images | Object classification in segmented satellite images | 96.5 |
|
|