Wilms tumor versus non-Wilms tumor, Oncology.
| Article DOI | https://doi.org/10.21037/tp-23-508 |
| Objective | To classify CT images for identification of Wilms and non-Wilms tumors. |
| AI Approach | SVM, RF, LDA, Adaboost, Gaussian process, Auto-Encoder, LR-Lasso, Decision tree, Naïve Bayes. |
| Data Source(s) | Single institutional series (82 patients) |
| Model Input | CT images (texture features) and clinical features (age, sex) |
| Model Outcome | Wilms tumor vs. non-Wilms tumor |
| Model Metrics | Corticomedullary phase: Validation set highest AUC: 0.79 Training Set AUC: 0.81 Testing Set AUC: 0.43 Nephrogenic phase: Validation set highest AUC: 0.66 Training Set AUC: 0.78 Testing Set AUC: 0.69 |
| Model Usability | NA |
AI = Artificial intelligence, SVM = Support Vector machines, RF = Random forrest, LR = Logistic Regression
AIPEDURO 
Leave a comment