悪性転化を伴う子宮内膜症におけるMRイメージングIVIMおよびT2マッピングの予測値

This study aims to explore the value of MRI intravoxel incoherent motion (IVIM) combined with T2 mapping in predicting the malignancy of endometriosis (EM) and construct a noninvasive preoperative risk assessment model. A retrospective analysis was conducted on 156 patients with pathologically confirmed EM or endometriosis - associated ovarian cancer, who were divided into the benign group (102 cases) and the malignant transformation group (54 cases). Two observers independently measured the IVIM parameters (D, D*, f) and T2 values to construct single-parameter and multi-parameter joint models. The receiver operator characteristic curve was used to evaluate the diagnostic efficacy, and logistic regression was used to analyze the independent predictors. The values of D, f, and T2 in the malignant transformation group were significantly lower than those in the benign group (all P < .05), while there was no statistically significant difference in D* value. The area under curve (AUC) of D, f, T2, and the combined model D + f + T2 were 0.737, 0.701, 0.773, and 0.874, respectively. The optimal cutoff values of D, f, and T2 were 1.10 × 10⁻³ mm2/s, 37.30 %, and 119.65 ms, respectively. Values lower than these may indicate a risk of malignant transformation of EM. Logistic regression confirmed that D, D*, and T2 were independent predictors of EM malignancy. IVIM combined with T2 mapping can non-invasively and quantitatively assess the risk of malignant transformation of EM. The combined model of parameters of two sequences has good potential for clinical promotion.