Abstract
Differentiation of uterine leiomyosarcoma (uLMS) from a benign uterine leiomyoma is one of the biggest problems in terms of diagnostics because of the similarities in the clinical picture and the radiographic characteristics, which can occur in both diseases. Undiagnosed uLMS before the operation can delay the correct therapy and cause possible complications of some operations, such as tumor morcellation. Recent advances in imaging, radiomics, artificial intelligence (AI), and biomarkers have therefore been regarded as promising ways to enhance diagnostic capabilities. This review aimed to summarize the current evidence on imaging modalities, radiomics, AI, and biomarkers for the preoperative detection of uterine leiomyosarcoma. A literature search was performed in PubMed, Scopus, and Google Scholar. English articles published from 2015 to 2025 and peer-reviewed articles presenting original research or review data on imaging, radiomics, AI, and biomarkers for Ulms were selected based on specific keywords. Multiparametric magnetic resonance imaging (MRI) still appears to be the most useful imaging modality. Radiomics and AI have revealed a promising diagnostic ability owing to enhanced lesion characterization and predictive models. Additionally, various serum and molecular biomarkers, such as lactate dehydrogenase (LDH) isoenzymes, inflammatory markers, and circulating nucleic acids, have been found to have a complementary diagnostic value. However, available data are limited due to the small sample size and lack of methodological homogeneity. The combination of imaging with radiomics, AI, and molecular biomarkers can contribute to preoperative uLMS diagnosis. Future multicentric prospective studies are required to prove their efficacy and validate them.

Key words: Uterine leiomyosarcoma, preoperative diagnosis, MRI, radiomics, artificial intelligence, biomarkers, review article