乳腺X线直方图鉴别BI-RADS 4类肿块良恶性的价值

Value of X-ray histogram in differentiating benign and malignant in breast masses of BI-RADS category 4

目的研究乳腺X线直方图分析对乳腺影像报告和数据系统(BI-RADS)4类肿块良、恶性鉴别诊断的价值。方法回顾性分析郑州大学第三附属医院经病理证实且乳腺X线图像被BI-RADS分类为4类的乳腺肿块患者138例,其中良性患者74例(肿块82枚)、恶性患者64例(肿块67枚)。分别在良、恶性两组肿块的乳腺X线图像上用软件MaZda(4.6)勾画肿块的感兴趣区(ROI),进行灰度直方图分析,对获得的两组直方图参数进行统计分析,比较各参数的统计学意义;并运用受试者工作曲线(ROC)评价有统计学意义的参数对BI-RADS 4类肿块良、恶性的鉴别诊断效能。结果直方图分析得到的9个纹理参数中,平均值、方差、第1、10、50、90、99百分数等7个参数差异有统计学意义(P<0.05),ROC曲线下的最大面积(AUC)分别为0.69、0.63、0.63、0.65、0.69、0.72、0.73,其中第99百分位数的AUC最大,取最佳阈值0.44时的特异度为56.1%、敏感度为80.6%;偏度、峰度2个参数差异无统计学意义(P>0.05)。结论X线直方图分析可以提供更多量化信息特征,为鉴别BI-RADS 4类乳腺肿块的良恶性提供了一定的参考价值。

【Objective】To explore the value of the histogram analysis derived from mammogram imaging in differentiating benign from malignant breast masses of BI-RADS category 4.【Methods】A retrospective analysis was made of 138 patients with BI-RADS 4 breast masses confirmed by operation and pathology in our hospital.Among them,74 were benign and 64 were malignant.MaZda 4.6 software was used to extract the region of interest(ROI)of the masses.Histogram analysis was carried out for ROI of the masses.Mean,variance,skewness,kurtosis,1 st,10 th,50 th,90 th and 99 th percentile were used to find out the significant difference between the benign and malignant masses.【Results】Among the nine texture parameters extracted from histogram,the average,variance,1,10,50,90 and 99 percentiles had statistical significance(P<0.05),and the ROC curve was drawn.The maximum area under the ROC curve was 0.69,0.63,0.63,0.65,0.69,0.72,0.73 respectively.The AUC of the ninety-ninth percentile was the largest,the specificity and sensitivity of the optimal threshold of 0.44 were 56.1%and 80.6%respectively.The skewness and kurtosis had no statistical significance(P>0.05).【Conclusion】Histogram analysis can provide more quantitative information features,and provide a certain reference value for the differentiation of benign and malignant in breast masses of BI-RADS category 4.