Table 3.
Head and neck.
| Publication | Year | Purpose | Image Modality | Patients | Ground Truth | Key Innovation | Outcome |
|---|---|---|---|---|---|---|---|
| Groendahl et al. (55) | 2021 | Tumors and involved nodes segmentation | PET/CT | 197 | Manual GTV delineation | Comparison of Traditional Methods with Machine Learning and Deep Learning Approaches | Dice = 0.75 |
| Taku et al. (56) | 2022 | Involved LN segmentation | CT | 110 | Pathological Report | Use DL-CNN to segment GTVT for HPV-associated OPC | Dice = 0.92 |
| Tekchandani et al. (57) | 2022 | LN segmentation and metastatic analysis |
CT | 175 | Pathological Report | Introduced LNdtnNet: combines attention, residual UNet, squeeze-excitation for CT | Dice = 0.94 |
| Wu et al. (58) | 2022 | Metastatic LN detection | CT | 114 | Pathologically Confirmed | Integrated lymph node features with GCN for top performance. | mFROC = 0.63 |
| Bollen et al. (59) | 2023 | GTV segmentation | CT/PET/MRI | 170 | GTV delineation | Developed automated GTV delineation using 3D CNN with multimodal imaging | mDice = 0.89 |
| Ariji et al. (60) | 2022 | Metastatic cervical LN segmentation | CT | 158 | Histopathological diagnosis | Utilize the U-Net architecture to segment metastatic cervical lymph nodes | F1 = 0.83 |
| Liao et al. (61) | 2025 | Involved LN segmentation | CT | 626 | Manual segmentation | Used a two-stage transfer learning approach with nnUNet pretraining | Dice = 0.72 |
| Al Hasan et al. (62) | 2024 | Small normal lymph nodes segmentation | CT | 221 | Manual segmentation | Enhanced Attentional U- Net by filtering encoder features for segmentation | Dice = 0.81 |