Table 3 ML-classifiers and DenseNet-121 performance on the independent test set
From: Tumor classification of gastrointestinal liver metastases using CT-based radiomics and deep learning
Random undersampling | SMOTE | ||||
|---|---|---|---|---|---|
Approach | Classifier | AUC | Accuracy | AUC | Accuracy |
ML-classifier | XG Boost | 0.71 | 0.84 | 0.71 | 0.82 |
Random Forest | 0.60 | 0.84 | 0.62 | 0.79 | |
K-means clustering SVM | 0.79 | 0.72 | 0.79 | 0.71 | |
K-nearest neighbour | 0.77 | 0.59 | 0.87 | 0.67 | |
SVM | 0.79 | 0.72 | 0.79 | 0.71 | |
Logistic Regression | 0.66 | 0.53 | 0.65 | 0.60 | |
Gaussian Naive Bayes | 0.51 | 0.52 | 0.52 | 0.50 | |
Decision Tree | 0.43 | 0.36 | 0.43 | 0.29 | |
CNN-classifier | DenseNet-121 | 0.80 | 0.83 | 0.80 | 0.83 |