磁共振扩散峰度成像鉴别不同腮腺疾病及诊断腮腺腺淋巴瘤的应用

Application of diffusion kurtosis imaging in differential diagnosis of parotid gland disease and diagnosis of parotid adenolymphoma

目的探讨磁共振扩散峰度成像(DKI)鉴别不同类型腮腺疾病及诊断腮腺腺淋巴瘤(PAL)的价值。方法回顾性分析57例腮腺疾病患者的DKI及DWI资料,分为感染性病变组(n=10)、混合瘤组(n=19)、PAL组(n=14)、其他良性肿瘤组(n=4)、恶性肿瘤组(n=10)。并将其中19例单侧腮腺病变患者的对侧正常腮腺作为正常对照组。比较各组病灶DKI扩散峰度系数(K_(mean)、K_(rad)、K_(ax))、扩散系数(D_(mean)、D_(rad)、D_(ax))、FA值及传统ADC值的差异。采用二分类Logistic回归筛选在PAL的诊断中具有统计学意义的指标,建立Logistic回归方程。绘制ROC曲线对筛选后的指标及二分类Logistic回归模型的诊断效能进行分析。结果各组间K_(mean)、K_(rad)、K_(ax)、D_(mean)、D_(rad)、D_(ax)、FA及ADC值的差别均有统计学意义(P均<0.05)。ROC曲线分析显示,FA联合K_(ax)值诊断PLA的曲线下面积(AUC)为0.88±0.06(0.79~0.94),高于单纯K_(ax)[0.80±0.07(0.70~0.88)]和FA的AUC[0.63±0.10(0.52~0.73)],差异均有统计学意义(P均<0.05);其敏感度、特异度、准确率、阳性预测值、阴性预测值分别为71.43%、95.78%、91.77%、76.92%、94.44%。结论 DKI可用于鉴别不同类型腮腺疾病,联合应用FA及K_(ax)值有利于提高对PAL的诊断能力。

Objective To investigate the value of diffusion kurtosis imaging （DKI） in differential diagnosis of parotid gland disease and diagnosis of parotid adenolymphoma （PAL）. Methods DKI and DWI data of 57 patients with parotid gland disease were etrospectively analyzed. Totally 57 cases were divided into infectious lesions group （n=10）, pleomorphic adenoma group （n=19）, PAL group （n=14）, other benign parotid tumor group （n=4） and malignant parotid tumor group （n=10）. Contralateral normal parotid glands in 19 patients with unilateral parotid gland lesions were treated as control group. The quantitative parameters including kurtosis concerning parameters （Kmean, Krad, Kax）, diffusivity concerning parameters （Dmean, Drad, Dax）, fractional anisotropy （FA） and conventional apparent diffusion coefficient （ADC） values were retrospectively reviewed. The binary Logistic regression method was used to confirm parameters with significant difference in diagnosing PAL. And Logistic regression equation was constructed to diagnose PAL. ROC analysis was conducted to evaluate the diagnostic value of the confirmed parameters and the Logistic regression equation. Results Significant difference of the parameters including Kmean, Krad, Kax, Dmean, Drad, Dax, FA and ADC values were found among different groups （all P〈0.05）. ROC analysis demonstrated a higher area under the curve （AUC） for FA＋Kax[0.88±0.06（0.79-0.94）] than Kax[0.80±0.07（0.70-0.88）] and FA[0.63±0.10（0.52-0.73）], respectively （both P〈0.05）. The sensitivity, specificity, accuracy, positive predictive value and negative predictive value was 71.43%, 95.78%, 91.77%, 76.92% and 94.44%. Conclusion DKI showed high diagnostic capacity in differential diagnosis of parotid gland disease. The combination of FA and Kax can improve the diagnostic accuracy in diagnosis of PAL.