摘要
目的:探讨多层螺旋CT图像纹理分析对卵巢癌转移性淋巴结的诊断价值。方法:回顾性收集2016年3月至2017年3月卵巢癌患者38例,所有患者均进行CT常规平扫和增强扫描。根据病理结果进行分组,将相应CT图像上区域淋巴结分为转移性淋巴结(metastatic lymph node,MLN)和非转移性淋巴结(non-metastatic lymph node,NLN)。测量并计算目标淋巴结的短径、短径/长径之比,通过分析门静脉期图像获得纹理分析参数(偏度、峰度、方差、熵和逆差矩)。采用独立样本t检验、Mann-Whitney U秩和检验、Logistic回归分析CT图像纹理分析对卵巢癌转移性淋巴结的诊断价值。结果:根据病理结果,38例卵巢癌患者术前CT图像上共选取淋巴结72个,其中MNL 37个,NLN 35个。卵巢癌MLN短径大于NLN,差异有统计学意义,但短径/长径之比在两组间无差异;卵巢癌MLN熵高于NLN,差异有统计学意义,而其他纹理特征参数并无统计学差异。淋巴结短径、熵的ROC下面积分别为0.74和0.83,熵的鉴别效能优于短径;短径和熵的诊断敏感度和特异度分别为78.38%、62.86%和78.38%、85.71%。以淋巴结有无转移为因变量,纳入淋巴结图像纹理特征参数,采用逐步回归筛选自变量,熵最终入选回归模型(OR=9.08,95%CI 3.32-21.35,P<0.05)。结论:多层螺旋CT图像纹理分析对于术前卵巢癌患者转移性淋巴结诊断具有重要意义,尤其是能够反映异质性的参数熵,能够为卵巢癌淋巴结N分期提供比较精准的依据。
Objective:To investigate the value of multi-slice spiral CT image texture analysis in differentiating metastatic lymph nodes(MLN)from non-metastatic lymph nodes(NLN)in patients with ovarian cancer.Methods:Thirty-eight patients of ovarian cancer patients in Wuwei People’s Hospital were retrospectively collected from August 2016 to March 2017.According to the pathological results,the regional lymph nodes in the corresponding CT images were divided into metastatic lymph nodes(MLN)and non-metastatic lymph nodes(NLN).The short-axis diameter and short to long-axis diameter ratio of lymph nodes were manually measured and calculated,and the texture analysis parameters(skewness,kurtosis,variance,inverse difference moment)were acquired.Independent sample t test,Mann-Whitney U rank sum test,Logistic were used to investigate the value of multi-slice spiral CT image texture analysis in differentiating MLN from NLN in patients with ovarian cancer.P<0.05 showed a statistically significant difference.Results:According to the pathological results,a total of 72 lymph nodes were selected from the preoperative CT images of 38 breast cancer patients,including 37 MNL and 35 NLN.The short diameter of MLN in ovarian cancer was greater than that of NLN,and the difference was statistically significant.Entropy of ovarian cancer in MLN were higher than in NLN,and the difference was statistically significant,while other texture feature parameters were not statistically significant.The area under ROC of short diameter and entropy of lymph nodes was 0.74 and 0.83,respectively.The diagnostic sensitivity and specificity of short diameter and entropy were 78.38%,62.86%and 78.38%,85.71%,respectively.Multivariate logistic regression analysis showed that only entropy(OR=9.08,95%CI 3.32~21.35,P<0.05)was screened out as the independent variable.Conclusion:Multi-slice spiral CT image texture analysis can facilitate the accurate differentiation between MLN and NLN in patients with ovarian cancer,and especially the entropy has the optimal reference significance.
作者
于志鹏
孟存忠
赵长秀
陈彬
赵小芳
李文荣
Yu Zhipeng;Meng Cunzhong;Zhao Changxiu;Chen Bin;Zhao Xiaofang;Li Wenrong(Department of Radiation Oncology,Wuwei People's Hospital,Gansu Wuwei 733000,China;Department of Imaging,the First Affiliated Hospital of Xi'an Jiaotong University,Shaanxi Xi'an 710061,China)
出处
《现代肿瘤医学》
CAS
2020年第18期3222-3226,共5页
Journal of Modern Oncology
关键词
卵巢癌
淋巴结转移
多层螺旋CT
纹理分析
ovarian cancer
lymphatic metastasis
multi-slice spiral CT
texture analysis
