Table 1 Quantitative comparison of state-of-the-art methods.
From: Multimodal deep learning for cephalometric landmark detection and treatment prediction
Category | Method | Year | Modality | Dataset size | Mean detection error (mm) | 2 mm success Rate (%) | Treatment prediction accuracy (%) |
|---|---|---|---|---|---|---|---|
Traditional | ASM25 | 1995 | Ceph | 247 | 3.82 | 51.2 | N/A |
Traditional | AAM25 | 1995 | Ceph | 247 | 3.44 | 54.8 | N/A |
Traditional | RFRV13 | 2015 | Ceph | 400 | 2.51 | 68.6 | N/A |
Deep learning | U-Net89 | 2015 | Ceph | 400 | 2.35 | 72.1 | N/A |
Deep learning | ResNet-based90 | 2020 | Ceph | 400 | 1.87 | 78.3 | N/A |
Deep learning | CephaNet33 | 2020 | Ceph | 735 | 1.58 | 85.2 | N/A |
Deep learning | AGNET93 | 2022 | Ceph + CBCT | 250 | 1.39 | 88.9 | N/A |
Treatment prediction | Conventional95 | 2016 | Ceph | 320 | N/A | N/A | 69.2 |
Treatment prediction | CNN-based107 | 2021 | Ceph | 450 | N/A | N/A | 73.5 |
Treatment prediction | 3D-based106 | 2020 | CBCT | 215 | N/A | N/A | 78.2 |
