Development and validation of a multi-parametric MRI diagnostic model for differentiating hemangioma-like metastases from small (< 3 cm) hepatic hemangiomas: a size-based subgroup analysis.

backgroundHemangioma-like metastases (HM) are rare but treacherous hypervascular malignancies that mimic the imaging features of benign hepatic hemangiomas (HH), particularly when lesions are small (< 3 cm). This resemblance creates a "diagnostic grey zone," often leading to misdiagnosis and inappropriate treatment delays. This study aims to develop and evaluate a multi-parametric MRI model to accurately distinguish small HM from HH and assess its diagnostic performance across different lesion size subgroups.

methodsThis retrospective study analyzed 149 lesions (81 HMs in 37 patients and 68 HHs in 48 patients), all smaller than 3 cm. Qualitative features on T2-weighted imaging (T2WI), diffusion-weighted imaging (DWI), and dynamic contrast-enhanced (DCE) MRI were systematically evaluated. Multivariate logistic regression was employed to identify independent predictors and construct a combined diagnostic model. The calibration of the nomogram was assessed using calibration plots with 1,000 bootstrap resamples. Decision curve analysis (DCA) was performed to evaluate the clinical utility of the model by quantifying the net benefit at different threshold probabilities. The diagnostic performance was further validated in three size-based subgroups: 5-<10 mm, 10-<20 mm, and 20-<30 mm.

resultsFour MRI features emerged as robust independent predictors of HM: moderately hyperintense T2WI signal (OR, 18.97; 95% CI, 3.65-98.72), heterogeneous T2WI architecture (OR, 54.28; 95% CI, 6.50-453.56), ring-like arterial enhancement (OR, 90.76; 95% CI, 10.61-776.68), and unclear delayed phase boundary (OR, 11.05; 95% CI, 1.50-81.65). The combined multi-parametric model achieved superior diagnostic performance compared to any single feature, yielding an area under the curve (AUC) of 0.981, with a sensitivity of 95.1% and a specificity of 97.1%. Subgroup analysis revealed that while the diagnostic accuracy of individual features (especially DWI) improved with increasing lesion size, the combined model maintained high diagnostic stability even in the challenging 5-<10 mm subgroup.

conclusionThe proposed multi-parametric MRI model offers an effective, non-invasive tool for differentiating small HMs from HHs. The presence of heterogeneous T2WI signal and ring-like arterial enhancement should trigger high suspicion of malignancy, even in sub-centimeter lesions. This model has the potential to aid clinicians in risk stratification and may help reduce unnecessary biopsies.