磁共振扩散加权成像联合T2WI图纹理分析定量评估乳腺良恶性结节的价值

目的：探讨磁共振扩散加权成像表观扩散系数（ADC）、T2加权成像（T2WI）图纹理参数定量评估乳腺良恶性结节的价值。方法：回顾性分析丽水市中心医院82例经病理证实为恶性或良性肿瘤的乳腺结节患者的磁共振ADC和T2WI图像，利用后处理软件对乳腺结节勾画感兴趣区（ROI），提取ADC图和T2WI纹理特征，生成多个一阶和二阶纹理参数。采用双样本t检验或Mann-Whitney U检验比较良恶性结节组纹理参数值差异。对差异有统计学意义的参数绘制ROC曲线，计算相应临界值AUC、敏感度和特异度，并用Z检验比较ADC图、T2WI纹理参数及二者联合的ROC曲线结果的差异。结果：良性结节组ADCmean、ADCmedian、ADCmin明显高于恶性结节组（均P＜0.05）。恶性结节组ADC图偏度值、熵值明显高于良性结节组（均P＜0.05）。ROC曲线显示，确定乳腺良恶性结节的ADCmean、ADCmedian、ADCmin、偏度值、熵值诊断阈值分别为1.34×10-3 mm2/s、1.41×10-3 mm2/s、0.97×10-3 mm2/s、-0.075、3.762，AUC分别为0.90（95%CI=0.82～0.95）、0.89（95%CI=0.80～0.95）、0.91（95%CI=0.83～0.96）、0.74（95%CI=0.64～0.83）、0.95（95%CI=0.87～0.98）；对ROC曲线比较显示熵值有最佳诊断效能，相应阈值的敏感度和特异度分别为80.55%、94.74%，比较ADC图、T2WI及二者联合的熵值AUC，显示二者联合的AUC值最大（AUC=0.97）。结论：基于ADC图和T2WI的纹理分析参数对鉴别乳腺良恶性结节有较高的诊断价值，尤以二阶纹理参数熵值诊断价值较大。

Quantitative evaluation of benign and malignant nodules of breast by diffusion-weighted imaging combined with T2WI texture analysis

Objective: To investigate the value of apparent diffusion coefficient (ADC) and T2-weighted imaging (T2WI) texture parameters in quantitative evaluation of benign and malignant breast nodules. Methods: The retrospective study was based on breast ADC and T2WI images. A total of 82 patients with histopathologically confirmed malignant or benign breast nodules from Lishui Municipal Central Hospital were enrolled in this study. Texture features from region of interest (ROI) of breast nodules were extracted from ADC map and T2WI images. Several first-order and second-order texture parameters were generated automatically by post-processing software. The difference of texture parameters between the two groups were compared by Student’s t-test or Mann-Whitney U test. The receiver operating characteristic curves (ROCs) were generated based on the significant variables identified from the univariate analysis, and the area under the curve (AUC), sensitivity, and specificity for nodules differentiation were reported. Z test was used for multiple comparisons of the AUCs with statistical differences between malignant and benign nodules groups. Results: ADCmean, ADCmedian and ADCmin values in breast benign nodule group were significantly higher than those in malignant nodule group (all P<0.05). The skewness and entropy of breast malignant nodule group were significantly higher than benign nodule group (all P<0.05). ROC showed that the thresholds of ADCmean, ADCmedian, ADCmin, skewness and entropy were 1.34×10-3 mm2/s, 1.41×10-3 mm2/s, 0.97×10-3 mm2/s, -0.075 and 3.762, respectively; and AUCs were 0.90 (95%CI= 0.82-0.95), 0.89 (95%CI=0.80-0.95), 0.91(95%CI=0.83-0.96), 0.74 (95%CI=0.64-0.83), 0.95 (95%CI=0.87-0.98), respectively. Comparison of ROCs showed that entropy had the best diagnostic efficiency, and the sensitivity and specificity were 80.55%, 94.74%. AUCs of entropy from ADC, T2WI and both combination showed that the ADC combined with T2WI had the largest entropy AUC (AUC=0.97). Conclusion: Texture parameters based on ADC combined with T2WI images can provide useful proof for differentiating benign breast nodules from malignant, especially for the second-order texture parameters entropy which is expected to provide more diagnostic value for tumor imaging in clinical practice.