MRI-based Intratumoral and Peritumoral Radiomics for Assessing Deep Myometrial Invasion with Early-Stage Endometrioid Adenocarcinoma

Motivation: Accurate assessment of deep myometrial invasion (DMI) in endometrial cancer (EC) is crucial for treatment planning and prognosis. Goal(s): This study aimed to develop radiomics models utilizing both intratumoral and peritumoral features to predict DMI in early-stage EC. Approach: A cohort of 459 patients underwent magnetic resonance imaging (MRI), and radiomics features were extracted from various imaging sequences. Multiple logistic regression models were constructed and validated. Results: The multimodal MRI-based radiomics model from the intratumoral and peritumoral regions expanded by 5 mm proved optimal, with the stacking and ensemble models demonstrating similar performance at 5 mm expansion. Impact: MRI-based radiomics models effectively predict deep myometrial invasion (DMI) in early-stage endometrioid adenocarcinoma (EAC). The multimodal radiomics model from the intratumoral and peritumoral regions expanded by 5 mm offers improved performance for detecting DMI in early-stage EAC.