Table 3 Previously published classification results of COPD versus non-COPD datasets.
From: A 3D-CNN model with CT-based parametric response mapping for classifying COPD subjects
Related work | Data | Methods | Accuracy (%) |
|---|---|---|---|
González et al.17 | Original CT slices of COPDGene testing cohort (n = 1000) | 2D-CNN | 77.3 |
Ran Du et al.18 | Multi-view snapshots of 3D lung-airway tree (190 COPD—90 Non-COPD) | 2D-CNN | 88.6 |
Ran Du et al.18 | 3D airway trees (190 COPD—90 Non-COPD) | 3D-CNN | 78.6 |
Feragen et al.19 | Airway trees (980 COPD and 986 Non-COPD subjects) | SVM | 64.9 |
Xu et al.27 | 1/6 of the total height (z) of the original CT sequences (190 COPD and 90 healthy control subjects) | Deep CNN transferred multiple instance learning (DCT-MIL) | 99.3 |
Our work | 3D PRM | 3D-CNN | 89.3 |
