Multisequence MRI‑based radiomics models for the assessment of microsatellite instability in endometrial cancer：A Feasibility Study

Motivation: Accurately evaluating microsatellite instability (MSI) status in endometrial cancer is crucial for optimizing immunotherapy strategies and improving patient outcomes. Goal(s): This study aims to assess the feasibility of using multiple machine learning models to predict MSI status in endometrial cancer (EC) and to identify the model with the best diagnostic performance. Approach: We assessed the diagnostic performance of the DT, SVM, KNN, LR, RF, EBM, and XGB models by calculating their respective area under the curve (AUC) values. Results: The XGB model achieved the highest MSI predictive performance, with notable specificity and positive predictive value, offering strong clinical utility for identifying high-risk patients. Impact: This study demonstrated the value of various radiomics models for non-invasively assessing MSI status prediction, providing reliable support for diagnosing and treating endometrial cancer (EC) and enhancing patient prognosis.