Table 1.
Summary of brain tumor diagnosis models.
| References | Feature extraction algorithm | Dataset | Classification models | No. of classes | Metrics |
|---|---|---|---|---|---|
| 1 | VGG-19 CNN | BTD, Radiopaedia | VGG-19 CNN | Three classes | Sensitivity 88.41%, specificity 96.12%, accuracy 94.51% |
| 4 | CNN, LSTM | BTDS, MBNDS, BMIDS | CNN-LSTM | Three classes | 99.22% accuracy |
| 5 | Histogram equalization, erosion, dilation | BTD | CNN | Two classes | Training recall 98.55%, validation recall 99.73% |
| 16 | CNN | BTDS | ResNet-CNN | Three classes | 99.90% accuracy, 100% specificity, 89.60% sensitivity |
| 6 | ELMA | BTD | FAHS-SVM | Two classes | 98.51% accuracy |
| 17 | – | MRI and SPECT images | KNN, SVM | Two classes | 96.8% accuracy, 95%, precision, 91%, f1-score |
| 8 | PCA, DWT | MRI images | SVM | Two classes | 99% accuracy |
| 18 | CNN | MRI images | Resnet50, InceptionV3 and MobileNetV2 | Three classes | 98.47%, 95.41%, 92.36% accuracy of models Resnet50, MobileNetV2 and InceptionV3 |
| 19 | DNN | MRI images | H-DNN | Two classes | 98.7% accuracy |
| 20 | Xception and IRV2 | BraTS 2018 dataset | Two-channel DNN | Two classes | 93.69% accuracy |
| 11 | VGG-16, CNN | MRI images | CNN, VGG-16 | Two classes | CNN 93.36%, VGG-16 97.16% |
| 21 | Inception-v3, DenseNet201 | BD | Xception-Ensemble | Three classes | Inception-v3 94.34% accuracy , DenseNet201 94.34% accuracy |
| 13 | Xception | MRI images | Xception, ResNet50, InceptionV3, VGG16, MobileNet | Three classes | Xception, ResNet50, InceptionV3, VGG16, and MobileNe, obtained F1-score, 98.75%, 98.50%, 98.00%, 97.50%, and 97.25% respectively |
| 14 | DWA | RIDER | AE-DNN | Two classes | 96% accuracy |
| 15 | ACNN | MRI images | ACNN | Two classes | 96.7% accuracy |