MRI Radiomics Data Analysis for Differentiation between Malignant Mixed Müllerian Tumors and Endometrial Carcinoma

Area of Science:

• Gynecologic Oncology
• Radiology
• Medical Imaging Analysis

Background:

• Malignant mixed Müllerian tumors (MMMTs) and endometrial carcinoma (EC) are distinct gynecologic malignancies.
• Differentiating between MMMTs and EC can be challenging, impacting treatment strategies and patient outcomes.
• Quantitative imaging features, such as radiomics, offer potential for improved tumor characterization.

Purpose of the Study:

• To compare quantitative radiomics data between MMMTs and EC.
• To identify texture features associated with overall survival (OS).
• To evaluate the utility of radiomics in distinguishing between EC and MMMTs.

Main Methods:

• Retrospective analysis of 61 patients (36 EC, 25 MMMTs).
• Extraction and analysis of radiomic and gray-level co-occurrence matrix (GLCM) features.
• Statistical comparison using Wilcoxon Rank sum and Fisher's exact tests.
• Logistic regression with elastic net for feature selection and Cox regression for survival analysis.

Main Results:

• Skewness and tumor volume significantly differed between EC and MMMTs (p ≤ 0.05).
• Cluster shade range, angular variance of cluster shade, and sum of squares variance range predicted EC status (p ≤ 0.05).
• The '256 Angular Variance of Energy' texture feature independently predicted OS (p = 0.004).

Conclusions:

• Tumor volume and specific texture features can help distinguish between EC and MMMTs.
• Radiomic analysis, particularly texture features, shows promise in predicting patient outcomes (OS).
• Quantitative imaging metrics may enhance diagnostic accuracy and prognostic assessment in gynecologic cancers.