Table 2 Overview of model architectures, training data, and metrics results from selected papers
From: A review of deep learning for brain tumor analysis in MRI
Rerence | Model architecture name | Training data used | Test set results (Dice) |
|---|---|---|---|
Myronenko et al.34 | Asymmetrical U-Net | BraTS 2018 | WT: 88.39, TC: 81.54, ET: 76.64 |
Jiang et al.35 | Two-Stage Cascaded U-Net | BraTS 2019 | WT: 88.80, TC: 83.70, ET: 83.27 |
Isensee et al.33 | nnU-Net (no new-Net) | BraTS 2020 | WT: 88.95, TC: 85.06, ET: 82.03 |
Luu and Park36 | modified nnU-Net | BraTS 2021 | WT: 92.75, TC: 87.81, ET: 84.51 |
Zeineldin et al.14 | Ensemble: DeepSeg, nnU-Net, and DeepSCAN | BraTS 2022 | WT: 92.94, TC: 87.88, ET: 88.03 |
