Evaluate lymphovascular space invasion in endometrial cancer using diffusion-weighted imaging-based habitat imaging

Lymphovascular space invasion (LVSI) is a key prognostic factor influencing treatment decisions in endometrial cancer (EC). Here, we evaluate whether diffusion-weighted imaging (DWI)-based habitat imaging can noninvasively identify LVSI in EC. We retrospectively analyzed 101 EC patients who underwent preoperative multi-b-value DWI examination between December 2020 and October 2024. EC lesions were decoded into four habitats determined by unsupervised K-means clustering using true diffusion coefficient (D), perfusion fraction (f), and mean kurtosis (MK) maps, and the volume fractions of each habitat were quantified. LVSI-positive EC ( n = 22 ) exhibited significantly higher volume fractions of Habitat 1 ( H1 ) and lower volume fractions of Habitat 2 ( H2 ) compared to LVSI-negative cases ( p < 0.001 ). Logistic regression identified independent risk factors for LVSI, including H1, FIGO stage, and histologic grade. H1 demonstrated comparable diagnostic performance (AUC, 0.76 ; 95%CI: 0.67, 0.84 ) to pathological indicators, while achieving the highest sensitivity ( 81.82% ). Additionally, H1 correlated positively with tumor volume, while H2 correlated negatively with histologic grade. These findings suggest DWI-based habitat imaging could serve as a valuable preoperative tool for noninvasive LVSI assessment in EC.