Unlocking Tumor Aggressiveness in Endometrial Cancer: AI-Driven PET/CT Radiomics and Machine Learning for Prediction of High-Risk Tumor Histology

TL;DR

This study uses AI and PET/CT scans to predict aggressive endometrial cancer subtypes before surgery, aiming to improve personalized treatment planning.

Contribution

The novel contribution is the identification of PET radiomic biomarkers combined with machine learning for non-invasive endometrial cancer risk stratification.

Findings

Artificial Neural Networks and Random Forest models achieved moderate discrimination between low-risk and high-risk endometrial cancer subtypes.

NGTDM_Coarseness and SUVmin were identified as key features reflecting tumor heterogeneity and metabolic activity.

Combined feature sets showed significant patient-level classification differences, suggesting potential clinical utility.

Abstract

Our work integrates advanced imaging analytics and machine learning to enhance preoperative risk stratification in endometrial cancer, an urgent requirement for personalized surgical planning. Our research explores the prognostic utility of [18F]-FDG PET/CT-derived metabolic, volumetric, and radiomic features in risk stratification of high-risk versus low-risk histological subtypes of endometrial cancer. This manuscript provides: Insights into novel PET radiomic biomarkers for endometrial cancer stratification. Evaluation of the clinical applicability of radiomics combined with machine learning in preoperative EC assessment. A step forward in improving non-invasive preoperative assessments.

Purpose: Accurate preoperative risk stratification in endometrial cancer (EC) is essential for guiding surgical and therapeutic decisions. This study aimed to evaluate the discriminative performance of [18F]-FDG PET/CT-derived radiomic features combined with machine learning models for differentiating low-risk (LRH-EC) and high-risk histology (HRH-EC) subtypes. Methods: A total of 159 patients with histopathologically confirmed EC who underwent preoperative [18F]-FDG PET/CT were retrospectively analyzed. Radiomic features were extracted using LIFEx version 7.4.0 software following IBSI guidelines. After FDR correction and Pearson correlation–based redundancy reduction (|r| > 0.80), 16 radiomic features were retained for modeling. Three feature configurations (Conventional PET parameters, Radiomics16, and Combined) were evaluated. Machine learning models were developed using stratified 5-fold cross-validation. Model performance was assessed using AUC, accuracy, sensitivity, specificity, F1-score, Wilson confidence intervals, DeLong’s test, and McNemar’s test. Results: Artificial Neural Network (ANN) (AUC = 0.709) and Random Forest (RF) (AUC = 0.686) achieved the highest discriminative performance within the Radiomics16 feature set. No statistically significant superiority between algorithms or feature configurations was observed by DeLong analysis. However, McNemar’s test demonstrated significant patient-level classification differences for the Combined ANN model (p < 0.001). NGTDM_Coarseness and SUVmin emerged as the most influential features, reflecting tumor heterogeneity and metabolic activity. Conclusions: [18F]-FDG PET/CT-based radiomics combined with machine learning provides moderate yet consistent discrimination between LRH-EC and HRH-EC. While external validation is required, this approach may support noninvasive preoperative risk stratification in endometrial cancer.

Figures