Table 12 Comparative result of the proposed Xl-Tl model.
From: Brain tumor detection empowered with ensemble deep learning approaches from MRI scan images
Study | Year | Data Augmentation | Dataset | Ensemble used | Deep learning models | Performance metrics (accuracy, etc.) | Key contributions | Limitations |
|---|---|---|---|---|---|---|---|---|
2020 | No | Private | No | Alexnet, googlenet, vggnet | Accuracy: 96% | Data augmentation applied | No ensemble model, limited dataset | |
2021 | No | Private | No | Berkeley wavelet transformation (bwt) | Not specified | Watermarking technique for feature extraction | Small dataset, handcrafted features, slow | |
2020 | No | Brain tumor dataset | No | Cnn (vgg16, resnet-50, inception-v3) | Vgg16: 96%, resnet-50: 89%, inception-v3: 75% | Cnn-based classification | No ensemble models, lower accuracy | |
2022 | No | Private | No | Densenet | Accuracy: 95% | Deep learning applied | No data augmentation, lacks ensemble approach | |
2021 | No | Brain tumor dataset | Yes | Scnn, vgg16 | Accuracy: 97.77% | Addressed overfitting & imbalance dataset | Limited comparative analysis | |
2023 | No | Brain tumor dataset | Yes | Not specified | Accuracy: 98.4% | Applied ensemble learning | Lacks computational efficiency analysis | |
2024 | No | Brain tumor dataset | No | Svm, knn, decision tree, naïve bayes | Svm: 62%, knn & dt: 63%, naïve bayes: 60% | Classical ml approach | Low accuracy, no deep learning applied | |
2024 | No | Brain tumor dataset | Yes | Feature ensemble, stacking ensemble | Feature ensemble: 97.71%, stacking: 97.40% | Applied feature ensemble learning | No real-time efficiency evaluation | |
2024 | No | Brats dataset | No | Custom model | Accuracy: 98% | Proposed a new model | No ensemble learning | |
Proposed model (xl-tl) | 2024 | Yes | Brain tumor dataset | Yes | Ensemble (inceptionv3 + xception) | Training: 98.3%, testing: 98.5% | High accuracy, ensemble approach applied | Computational efficiency not analyzed |
