Related Experiment Video For benign endometrial lesions
Updated: Aug 6, 2026
Sentinel Lymph Node Mapping and Biopsy for Endometrial Cancer at Early Stage with Laparoscopy
Published on: August 19, 2021
Intratumoral and peritumoral MRI habitat imaging for differentiating stage IA endometrial cancer from benign
Yunzhu Wu1, Xianhong Wang2, Cheng Deng3
1Jiangsu Key Laboratory of Intelligent Medical Image Computing, School of Artificial Intelligence, Nanjing University of Information Science and Technology, Nanjing, China.
Purpose:
The aim of this study was to evaluate the value of different multiparametric MRI-based radiomics models in differentiating stage IA endometrial cancer (EC) from benign endometrial lesions.
Methods:
This retrospective study included 787 patients with stage IA endometrial cancer (EC) or benign endometrial lesions from four centers. Tumor regions of interest (ROIs) were manually delineated on MRI. Employing Python, the following peritumoral ROIs were automatically generated: 3-mm dilated and eroded peritumoral loops (LDE3), 3-mm eroded peritumoral loops (LE3), and intratumoral regions merged with 3-mm dilated peritumoral loops (RD3) Habitat clustering was performed using K-means and Gaussian Mixture Model (GMM) algorithms. Logistic regression was utilized to identify independent predictors and construct habitat-only and combined (clinical + habitat) models. Performance was evaluated using the area under the curve (AUC).
Results:
Age (P = 0.002) and vaginal bleeding (P < 0.001) were identified as independent clinical predictors of stage IA EC. Habitat-based models outperformed the clinical model in some external validation cohorts. Peritumoral features, particularly K-means_RD3, exhibited favorable robustness, achieving an average external validation AUC of 0.740. The integration of habitat features with clinical predictors yielded synergistic improvements, with the Clinical+K-means_RD3 model achieving the highest diagnostic performance (peak AUC: 0.921). Notably, this combined framework effectively mitigated the limitations of clinical factors in challenging subsets, elevating the AUC from 0.644 to 0.850 in validation group D. K-means demonstrated superior robustness and stability compared to GMM.
Conclusion:
Intratumoral and peritumoral habitat imaging based on multiparametric MRI can non-invasively reveal the microstructural characteristics of stage IA EC. Integrating clinical predictors with habitat models showed good diagnostic performance. The peritumoral habitat model exceeded the intratumoral habitat model.
