Table 3 The preprocessing, feature extraction, and feature selection method of deep learning method in COPD identification and stage
From: Artificial intelligence in COPD CT images: identification, staging, and quantitation
Team | Reference | Preprocessing | Feature extraction | Feature selection |
|---|---|---|---|---|
Mets et al. | [64] | The segmentation of lung and airway | Three quantitative CT biomarkers (emphysema, air trapping, and bronchial wall thickness) | - |
González et al. | [66] | Join four views into a single montage | CNN features | - |
Cheplygina et al. | [67] | 3D Region of interest (ROI) from CT image | Gaussian scale space features | - |
Sathiya et al. | [68] | Gray Scale | Gray Level Co-occurrence Matrix | - |
Xu et al. | [70] | The segmentation of lung from CT image | CNN features (AlexNet) | Principle component analysis |
Tang et al. | [72] | Lung mask generation, spatial normalisation | CNN features (ResNet-152) | - |
Hasenstab et al. | [74] | Co-registration, lung segmentation | Emphysema and air trapping feature | - |
Li et al. | [76] | Volume of Interest segmentation from CT | 1395 radiomics features | Variance threshold, Select K Best method, and least absolute shrinkage and selection operator (LASSO) |
Yang et al. | [80] | Lung region segmentation | 1316 radiomics features | LASSO |
Yang et al. | [81] | Lung parenchyma segmentation | 1316 radiomics features | Generalized linear model and LASSO |
Puchakayala et al. | [86] | Segmentation of lung and airways | Demographics features, emphysema feature, lung and airway radiomics features | - |