Work

Classification of uterine sarcoma and leiomyoma with quantitative CT and dual-energy iodine mapping

Year: 2026

Type: article

Abstract: PURPOSE: Preoperatively differentiating uterine sarcomas from leiomyomas remains challenging using routine imaging. We evaluated whether quantitative venous-phase CT metrics, such as attenuation profiling and dual-energy CT (DECT) iodine mapping, can aid in discrimination. METHODS: In this retrospective single-center study, cases of women with histopathology-proven leiomyoma or uterine sarcoma who underwent contrast-enhanced abdominopelvic CT (01/2010-01/2025) were included; lesions with dominant macrocalcifications were excluded. On venous-phase CT, ROI-based HU profiles were computed on the slice showing the maximal cross-sectional area of each lesion; the leading profile-derived parameters were the intralesional minimum HU, maximal HU, and the ΔHU (ΔHU = max-min). Reader-consensus values were analyzed. In a DECT subset of 20 cases, readers placed lesion and aortic ROIs to compute normalized iodine concentration (NIC = iodine_lesion/iodine_aorta). RESULTS: 81 patients were analyzed (leiomyoma n = 41, sarcoma n = 40; sarcoma subtypes: LMS n = 15, MMMT/CS n = 10, ESS n = 9, adenosarcoma n = 3, and UUS n = 3). Sarcomas showed lower minimum HU and higher ΔHU than leiomyomas (all p < 0.001), yielding AUC = 0.75 (minimum HU) and AUC = 0.73 (ΔHU). DECT with normalized iodine concentration (NIC) was available in n = 20 (10/10) and was higher in sarcomas, achieving AUC = 0.96 (Youden = 0.733, sensitivity = 0.80, specificity = 1.00). Reliability was good-to-excellent for HU features (ICC = 0.86-0.97) and excellent for iodine metrics (ICC = 0.99-1.00). CONCLUSION: Simple, aorta-normalized CT metrics may capture biologically plausible differences between uterine leiomyomas and sarcomas. In particular, aorta-normalized attenuation features showed fair discriminatory performance in the overall cohort, while DECT-derived NIC showed promising but exploratory performance in a small spectral subset that requires external confirmation.

Source: Clinical Imaging

DOI: 10.1016/j.clinimag.2026.110795