Table 3.
Nodule diagnosis accuracy between other methods and the classification network in our method on Luna16 dataset
| Diagnostician | Year | Accuracy |
|---|---|---|
| Vanilla 3D CNN [37] | 2016 | 87.40% |
| Multi-crop CNN [33] | 2017 | 87.14% |
| 3D DPN [39] | 2017 | 88.76% |
| DenseNet [7] | 2018 | 91.54% |
| SE-ResNeXt [21] | 2018 | 91.31% |
| Hierarchical semantic CNN [32] | 2019 | 90.82% |
| IDSNet [18] | 2020 | 91.10% |
| Our method | 2021 | 92.75% |
Bold text indicates the best results obtained in the comparative experiments