A Radiomic-Based Machine Learning Model Predicts Endometrial Cancer Recurrence Using Preoperative CT Radiomic Features: A Pilot Study

Abstract

Background: Current prognostic models lack the use of pre-operative CT images to predict recurrence in endometrial cancer (EC) patients. Our study aimed to investigate the potential of radiomic features extracted from pre-surgical CT scans to accurately predict disease-free survival (DFS) among EC patients.

Methods: Contrast-Enhanced CT (CE-CT) scans from 81 EC cases were used to extract the radiomic features from semi-automatically contoured volumes of interest. We employed a 10-fold cross-validation approach with a 6:4 training to test set and utilized data augmentation and balancing techniques. Univariate analysis was applied for feature reduction leading to the development of three distinct machine learning (ML) models for the prediction of DFS: LASSO-Cox, CoxBoost and Random Forest (RFsrc).

Results: In the training set, the ML models demonstrated AUCs ranging from 0.92 to 0.93, sensitivities from 0.96 to 1.00 and specificities from 0.77 to 0.89. In the test set, AUCs ranged from 0.86 to 0.90, sensitivities from 0.89 to 1.00 and specificities from 0.73 to 0.90. Patients classified as having a high recurrence risk prediction by ML models exhibited significantly worse DSF (p-value < 0.001) across all models.

Conclusions: Our findings demonstrate the potential of radiomics in predicting EC recurrence. While further validation studies are needed, our results underscore the promising role of radiomics in forecasting EC outcomes.

Keywords: artificial intelligence; endometrial cancer; personalized medicine; pre-surgical risk; prognosis; radiomic; recurrence.

Figure 1

Study design. CE-CT images were delineated, and Radiomic Features (RFs) were automatically extracted using Pyradiomics. Each patient was associated with observed disease-free survival. The dataset was divided into training and test sets with a 60:40 ratio 100 times. After the application of the data augmentation and balancing operations, the training data consisted of 318 EC patients of which 158 had recurrence (49.7%). A RFs reduction was performed using Kaplan and Meier curves and calculating the p-value of log-rank test assuming a cutoff of 0.20. Thus, the optimal model was applied using the following ML-based models: LASSO-Cox, CoxBoost and RFsrc.

Figure 2

CE-CT images in axial, coronal and sagittal view, and the 3D visualization of the contoured VOI from a patient with early EC recurrence (A) and without early recurrence during study follow-up (B). The images were acquired in the venous phase of contrast enhancement. CE-CT: contrast enhanced computer tomography, VOI: volume of interest; EC: endometrial cancer.

Figure 3

Performance of the radiomic models on the test EC cohort with the Kaplan-Meier curves showing patients’ recurrence based on the three ML models’ predictions; pred_low-risk and pred_high-risk: ML predictive scores for recurrence.