全容积ADC直方图分析鉴别乳腺病变良恶性的价值

Value of full volume apparent diffusion coefficient histogram analysis to the differential diagnosis of benign and malignant breast lesions

目的探讨MRI表观扩散系数(apparent diffusion coefficient, ADC)全容积直方图分析在乳腺良、恶性病变鉴别诊断中的价值。方法乳腺病变患者142例(156个病变),其中良性病变72例(81个病变)为良性病变组,恶性病变70例(75个病变)为恶性病变组。术前均行乳腺MRI检查,应用MaZda软件提取、分析病变的全容积ADC图像,通过ADC图像的像素值及灰度分布情况得到均值、方差、偏度、峰度及第1、10、50、90、99百分位数9个直方图参数,比较2组ADC全容积直方图参数;绘制ROC曲线,评估ADC全容积直方图各参数鉴别诊断乳腺病变良、恶性的效能。结果恶性病变组均值(119.05±16.58)、方差(383.68±216.37)及第1、10、50、90、99百分位数(78.56±20.24、96.19±17.18、117.93±16.57、143.77±19.45、167.05±24.65)均低于良性病变组(167.33±32.79、542.57±494.58、116.11±31.92、139.01±33.33、168.70±35.05、193.51±34.52、209.31±31.10)(P<0.05),偏度(0.30±0.64)高于良性病变组(-0.28±0.87)(P<0.05);均值和第50百分位数最佳截断值为0.69、0.72时,鉴别诊断乳腺良、恶性病变的AUC分别为0.910、0.912,灵敏度分别为84%、84%,特异度分别为85%和88%。结论 ADC全容积直方图参数鉴别诊断乳腺良、恶性病变具有较高的特异度和灵敏度,有一定鉴别诊断价值。

Objective To investigate the value of full volume apparent diffusion coefficient(ADC) histogram analysis to the differential diagnosis of benign and malignant breast diseases. Methods In 142 patients(156 lesions), there were 72 patients with 81 benign lesions(benign group) and 70 patients with 75 malignant lesions(malignant group). The MRI findings before operation were analyzed retrospectively. MaZda software was used to extract and analyze the full volume ADC images of lesions to obtain 9 histogram parameters as mean, variance, skewness, kurtosis, as well as the 1 st, 10 th, 50 th, 90 th and 99 th percentiles(Perc.1%, Perc.10%, Perc.50%, Perc.90%, Perc.99%). The histogram parameters of benign and malignant breast lesions were analyzed by univariate analysis.ROC was used to analyze the diagnostic efficiency of ADC full volume histogram parameters on the differential diagnosis of benign and malignant breast lesions. Results The mean(119.05±16.58), variance(383.68±216.37), Perc.1%(78.56±20.24), Perc.10%(96.19±17.18), Perc.50%(117.93±16.57), Perc.90%(143.77±19.45) and Perc.99%(167.05±24.65) in malignant group were lower than those in benign group(167.33±32.79, 542.57±494.58, 116.11±31.92, 139.01±33.33, 168.70±35.05, 193.51±34.52, 209.31±31.10)(P<0.05), while the skewness was higher in malignant group(0.30±0.64) than that in benign group(-0.28±0.87)(P<0.05). When the optimal cut-off values of mean and Perc.50% were 0.69 and 0.72, the AUCs for differential diagnosis was 0.910 and 0.912, with the sensitivities of 84% and 84%, and the specificities of 85% and 88%, respectively. Conclusion ADC full volume histogram parameters have high specificity and sensitivity in the differential diagnosis of benign and malignant breast lesions.