Developing a Radiomics and Machine Learning Model Applied to Magnetic Resonance Images for Differential Diagnosis of Myometrial Tumors
Authors:
- Valentina ChiappaVCValentina Chiappa
- Giulia GremmoGGGiulia Gremmo
- Matteo InterlenghiMatteo Interlenghi
- Christian SalvatoreCSChristian Salvatore
- Giorgio BoganiGBGiorgio Bogani
- Simona PalladinoSPSimona Palladino
- Umberto Leone Roberti MaggioreUMUmberto Leone Roberti Maggiore
- Giuseppina CalaresoGCGiuseppina Calareso
- Biagio PaoliniBPBiagio Paolini
- Lucia ZanchiLZLucia Zanchi
- Giulia BrugiaviniGBGiulia Brugiavini
- Francesco RaspagliesiFRFrancesco Raspagliesi
Preprint
- 11 March 2026
- preprint
- Published by MDPI AG in Preprints.org
Abstract
Background. The preoperative differential diagnosis of myometrial lesions remains a significant challenge when using conventional imaging techniques, such as ultrasound (US) and magnetic resonance imaging (MRI). Radiomics and machine learning, which leverage quantitative features beyond human visual perception, are increasingly recognized as promising tools for improving differential diagnosis in gynecology. Methods. This retrospective study included patients who underwent surgery for uterine masses and had preoperative MR. A machine learning model was developed to analyze radiomic features extracted from T2-weighted and diffusion-weighted MR images. Results. 44 subjects were included: 19 (43.2%) classified as "sarcoma" and 25 (56.8%) as "fibroid" based on histology after surgery. This dataset was used for training and cross-validation of different models. Three models, comprising ensembles of machine learning classifiers (random forests, support vector machines, and k-nearest neighbors), were developed for binary classification using histological diagnosis as reference standard. The best-performing model achieved the following results: AUC 90%, accuracy 82%, sensitivity 95%, specificity 72%, PPV 72%, and NPV 95%. Conclusions. Our model demonstrated high sensitivity and moderate accuracy, suggesting its potential as a valuable tool for assisting clinicians in the preliminary assessment of myometrial lesions and guiding decision-making toward conservative management in cases of non-suspicious masses.Keywords
All Related Versions
This publication has 25 references indexed in Scilit:
- MRI, clinical, and radiomic models for differentiation of uterine leiomyosarcoma and leiomyomaAbdominal Radiology, 2024
- Diagnostic accuracy of MRI in the differential diagnosis between uterine leiomyomas and sarcomas: A systematic review and meta‐analysisInternational Journal of Gynecology & Obstetrics, 2023
- Uterine mesenchymal tumors: development and preliminary results of a magnetic resonance imaging (MRI) diagnostic algorithmLa radiologia medica, 2023
- Combining multiparametric MRI features-based transfer learning and clinical parameters: application of machine learning for the differentiation of uterine sarcomas from atypical leiomyomasEuropean Radiology, 2022
- Machine Learning to Differentiate T2-Weighted Hyperintense Uterine Leiomyomas from Uterine Sarcomas by Utilizing Multiparametric Magnetic Resonance Quantitative Imaging FeaturesAcademic Radiology, 2019
- Imaging in gynecological disease (15): clinical and ultrasound characteristics of uterine sarcomaUltrasound in Obstetrics & Gynecology, 2019
- Combined use of T2-weighted and diffusion-weighted 3-T MR imaging for differentiating uterine sarcomas from benign leiomyomasEuropean Radiology, 2009
- The utility of diffusion-weighted MR imaging for differentiating uterine sarcomas from benign leiomyomasEuropean Radiology, 2007
- Can gray‐scale and color Doppler sonography differentiate between uterine leiomyosarcoma and leiomyoma?Journal of Clinical Ultrasound, 2007
- Uterine sarcomas versus leiomyomas: Gray‐scale and Doppler sonographic findingsJournal of Clinical Ultrasound, 2004