IVIM成像与DWI在乳腺良恶性病变鉴别诊断中的比较

目的 比较体素内不相干运动(IVIM)成像双指数模型、拉伸指数模型与扩散加权成像(DWI)单指数模型各参数在乳腺良恶性病变鉴别诊断中的价值.方法 回顾性分析257例经病理证实的乳腺病变患者(共276个病灶,包括197个恶性病变,79个良性病变).所有患者均行MRI常规检查及多b值DWI检查,获得传统DWI及IVIM各参数.比较各参数在正常乳腺组织、乳腺良性病变及恶性病变中的统计学差异,采用受试者工作特征(ROC)曲线确定各参数诊断乳腺恶性病变的阈值以及曲线下面积(AUC)、诊断敏感性和特异性.结果 正常乳腺组织、乳腺良性病变及恶性病变的表观扩散系数(ADC)、慢速表观扩散系数(slow ADC)、快速表观扩散系数(fast ADC)、灌注分数(f)、扩散分布指数(DDC)及扩散异质性指数(α)值均有统计学差异(P<0.001).ADC、slow ADC、f、DDC和α的AUC分别为0.865、0.861、0.742、0.85和0.735;ADC、slow ADC、DDC和α的最佳诊断阈值分别为1.105×10-3 mm2/s,0.883×10-3 mm2/s,1.025×10-3 mm2/s和0.842,slow ADC敏感性最高(90.3%),DDC特异性最高(79.5%).双指数模型中slow ADC与fast ADC联合诊断的AUC为0.882;拉伸指数模型DDC与α联合诊断的AUC为0.853.结论 3种模型对于乳腺病变良恶性的鉴别都具有较高的价值,传统ADC的诊断准确性较高,slow ADC敏感性较高,DDC特异性较高.双指数模型中slow ADC与fast ADC联合诊断具有较高的价值.

Comparison between the value of the IVIM imaging and DWI in differential diagnosis of benign and malignant breast lesions

Objective To compare the value of the intravoxel incoherent motion (IVIM)imaging (bi-exponential model and stretched-exponential model)and diffusion-weighted imaging (DWI)(mono-exponential model )in the differential diagnosis of benign and malignant breast lesions.Methods 257 patients with 276 breast lesions (including 197 malignant lesions and 79 benign lesions) confirmed by pathology were analyzed retrospectively.All of the patients underwent routine MRI sequences and multiple b-value DWI sequences to obtain the bi-exponential model,stretched-exponential model parameters of the IVIM and the mono-exponential model parameters of the traditional DWI.These parameters were compared among the normal breast tissues,benign lesions and malignant lesions.The sensitivity,specificity and best thresholds of these parameters for differential diagnosing malignant and benign lesions were calculated by the receiver operating characteristic (ROC)curve.Results The ADC,slow ADC,fast ADC,f,DDC andαvalues were significantly different among the normal breast tissues,benign and malignant lesions (P<0.001).The areas under the ROCcurve (AUC)of ADC,slow ADC,f,DDC andαvalues were 0 .865 , 0.861,0.742,0.85 and 0.735,respectively.The optimal diagnostic threshold of ADC,slow ADC,DDC andαvalues were 1.105× 10-3 mm2/s,0.883×10-3 mm2/s,1.025×10-3 mm2/s and0.842 respectively.The sensitivity is highest for slow ADC (90.3%),and the specificity is highest for DDC (79.5%).The AUC of combining slow ADC with fast ADC was 0.882.The AUC of combining DDC andαvalue was 0.853.Conclusion The parameters (ADC,slow ADC,fast ADC,f,DDC andα)obtained from the three models are all help to differential diagnosis of benign and malignant breast lesions.The ADC,slow ADC and DDC has the highest diagnostic accuracy,sensitivity and specificity,respectively.Combining slow ADC and fast ADC derived from the bi-exponential mode can provide a higher diagnostic accuracy.