Assessing whether deep learning super-resolution applied to contrast-enhanced MRI improves radiomic prediction of a common molecular subtype of endometrial cancer.
Key Takeaways
- Study used 140 surgically confirmed endometrial cancer patients split into training and testing cohorts
- Super-resolution DCE radiomic models achieved higher testing AUCs (up to 0.841) than original-resolution models
- Logistic regression and SVM on SR images significantly outperformed MLP in the testing set