中国癌症杂志 ›› 2018, Vol. 28 ›› Issue (3): 191-196.doi: 10.19401/j.cnki.1007-3639.2018.03.004

• 论著 • 上一篇    下一篇

肝脏磁共振T2WI图像纹理特征预测肝细胞癌患者微血管侵犯的价值

武明辉,谭红娜,吴青霞,史大鹏,王梅云   

  1. 河南省人民医院暨郑州大学人民医院放射科,河南 郑州 450003
  • 出版日期:2018-03-30 发布日期:2018-04-11
  • 通信作者: 王梅云 E-mail: marian9999@163.com

Value of MRI T2-weighted image texture analysis in evaluating the microvascular invasion for hepatocellular carcinoma

WU Minghui, TAN Hongna, WU Qingxia, SHI Dapeng, WANG Meiyun   

  1. Department of Radiology, Henan Provincial People’s Hospital & Zhengzhou University People’s Hospital, Zhengzhou 450003, Henan Province, China
  • Published:2018-03-30 Online:2018-04-11
  • Contact: WANG Meiyun E-mail: marian9999@163.com

摘要: 背景与目的:微血管侵犯(microvascular invasion,MVI)是肝细胞癌(hepatocellular carcinoma,HCC)术后肿瘤复发的重要危险因素之一,术前能否准确预测微血管侵犯会影响肿瘤治疗方案的选择,本研究旨在探讨肝脏MRI T2加权成像(T2-weighted imaging,T2WI)图像纹理特征预测HCC MVI的价值。方法:回顾性分析220例病理证实为HCC的T2WI图像及临床资料,并根据手术病理结果将患者分为MVI阳性组和MVI阴性组。分析两组HCC患者的T2WI图像纹理特征,应用纹理分析方法测定均值、标准差、偏度、峰度、方差、能量、熵、自相关、惯量、逆差距和反差共11个纹理参数,并对所得数据进行统计学分析。结果:220例HCC患者中,男性186例,女性34例,年龄27~84岁,中位年龄54岁;病理证实MVI阳性者71例,阴性者149例。220例HCC患者中,T2WI上检出MVI阳性患者93例,阴性患者127例,灵敏度和特异度分别为60.6%和66.4%。220例HCC患者T2WI图像纹理分析结果显示,MVI阳性组峰度、逆差距及自相关值均高于MVI阴性组,而惯量、熵值低于MVI阴性组,且这些纹理参数值差异均有统计学意义(P<0.05);其余纹理特征参数值差异无统计学意义(P>0.05)。峰度、熵、惯量、自相关和逆差距的曲线下面积(area under curve,AUC)分别为0.621、0.318、0.355、0.698和0.677。T2WI、纹理特征和T2WI联合纹理特征判断MVI的AUC分别为0.635、0.719和0.782;纹理特征和T2WI联合纹理特征诊断MVI的灵敏度分别为74.6%和88.1%,特异度分别为70.5%和74.5%。结论:T2WI图像纹理参数对HCC患者MVI有一定的预测作用,且T2WI联合纹理特征诊断MVI的效能高于单独T2WI及图像纹理分析。

关键词: 纹理分析, 肝细胞癌, 微血管浸润, MRI

Abstract: Background and purpose: Microvascular invasion (MVI) is one of the most important risk factors for postoperative recurrence of hepatocellular carcinoma (HCC). Accurate prediction of MVI before operation affects the selection of treatment strategy. The purpose of this study was to explore the value of MRI T2-weighted imaging (T2WI) texture analysis in evaluating the MVI for HCC. Methods: T2WI findings and clinical data were retrospectively reviewed in 220 patients with HCC confirmed by pathology, and all patients underwent surgical operation. Then these patients were divided into the with MVI group and the without MVI group according to the results of pathology. The texture features of these lesions were statistically analyzed, including mean value, standard deviation, skewness, kurtosis, variance, energy, entropy, correlation, interia, inverse difference moment and contrast. Results: A total of 220 HCC patients were enrolled in the study, including 186 males and 34 females, aged from 27 to 84 (median age 54). There were 71 cases with MVI confirmed by pathology. On T2WI of these 220 HCC patients, 93 cases showed with MVI while 127 patients did not have MVI; the sensitivity and specificity were 60.6% and 66.4%, respectively. The results of T2WI texture analysis for 220 HCC patients showed the values of kurtosis, inverse difference moment and correlation in the group with MVI were  higher compared to the group without MVI, while the values of entropy and interia in the group with MVI were lower compared to the group without MVI; there were significant differences in all these texture parameters between the two groups (P<0.05). However, there was no significant difference in the rest of texture parameters between the two groups (P>0.05). The values of area under curve (AUC) for texture parameters of kurtosis, entropy, interia, inverse difference moment and correlation were 0.621, 0.318, 0.355, 0.698 and 0.677, respectively. For the diagnostic efficiency in the with MVI group, the values of AUC using T2WI, T2WI texture features, and the combination of T2WI and texture features were 0.635, 0.719 and 0.782, respectively; the sensitivity and specificity of texture features and the combination of T2WI and texture features were 74.6% and 88.1%, 70.5% and 74.5%, respectively. Conclusion: T2WI texture parameters are helpful for predicting MVI, and the diagnostic efficiency of the combination of T2WI and texture features is higher than that of T2WI or texture analysis.

Key words: Texture analysis, Hepatocellular carcinoma, Microvascular invasion, MRI