Table 1.
State-of-art comparison.
| Ref./year | Dataset | Technique used | Performance parameters |
|---|---|---|---|
| 12/2018 | BRATS 2015 | Cascaded CNN | Dice coefficient = 0.89 |
| 13/2008 | Vivo brain tumors | Hybrid model | Jaccard Index = 0.69 |
| 14/2012 | Synthetic data from Utah + in vivo data from Harvard | Cellular automata model | Dice coefficient = 0.72 |
| 15/2013 | Web data + in vivo brain tumors | Lesion localization and segmentation model | Accuracy = 83–95% |
| 16/2016 | MICCAI-BRATS 2013 | SVM | Dice coefficient = 0.86 |
| 17/2017 | MICCAI-BRATS 2013 | Wavelet-based features | Dice coefficient = 0.88 |
| 18/2015 | MICCAI-BRATS 2013 | Random forest model | Dice coefficient = 0.88 |
| 19/2015 | MICCAI-BRATS 2013 | Using appearance- and context-based features | Dice coefficient = 0.83 |
| 20/2014 | MICCAI-BRATS 2013 | 3D CNN with 3D convolutional kernels | Dice coefficient = 0.87 |
| 39/2021 | BraTs2020 | U-Net with MobileNetV2 encoder | Dice Coefficient = 0.88 |
| Proposed |
Brain MRI, BraTs2020 |
LinkNet-34 semantic segmentation with EfficientNetB7 encoder |
Brain MRI Jaccard index = 0.89, Dice coefficient = 0.92 BraTs2020 Jaccard index = 0.75, Dice coefficient = 0.85 |