MR-T_(2)WI纹理分析对前列腺中央腺体低信号结节良恶性的鉴别

A Study of Texture Analysis based on Magnetic Resonance T;-Weighted Imaging in Differentiating Benign from Malignant Low Signal Nodules of the Prostatic Central Glandular

目的探讨MR-T_(2)WI纹理分析鉴别前列腺中央腺体低信号结节良恶性的价值。方法回顾性分析符合纳入标准的28例前列腺癌和80例前列腺增生患者多序列MRI检查,并经组织病理学证实。基于MR-T_(2)WI图像勾画三维兴趣区,使用纹理分析软件(Omni-Kinertics GE)提取ROI内一级灰度直方图及二阶纹理特征参数:能量、熵、偏度、峰值、均匀度、平均值和标准差。用统计学方法评估MR-T_(2)WI各特征纹理参数在前列腺癌、前列腺T_(2)WI低信号增生结节及前列腺T_(2)WI混杂偏低信号结节之间的差异。采用受试者工作特征曲线(ROC)和曲线下面积(AUC),评估各参数的诊断效能,并评价定量纹理参数对前列腺癌和前列腺T_(2)WI低信号结节的鉴别诊断效能。结果MR-T_(2)WI图像纹理分析定量参数中,能量、熵、偏度、均匀度和标准差5个参数在前列腺癌组与前列腺增生结节组间的差异具有统计学意义,峰值和平均值于各组间差异无统计学意义。标准差鉴别前列腺癌组与前列腺增生结节组的AUC值为0.832,当截点为0.5804时,其诊断敏感度为68.75%、特异度为89.29%,其效能最高;而熵则同时有较高的敏感度及诊断效能。结论MR-T_(2)WI纹理分析的部分定量参数有助于前列腺癌与良性前列腺增生T_(2)WI低信号病变的鉴别,为两者的鉴别诊断提供客观依据。

Objective To evaluate the values of texture analysis of magnetic resonance(MR)T_(2)-weighted imaging(T_(2)WI)in differentiating benign from malignant lesions in the central gland of prostate.Methods In this study,28 patients with prostate cancer and 80 patients with prostatic hyperplasia confirmed by histopathology were eligible and retrospectively analyzed with the multi-sequence MRI.The region of interest(ROI)for the lesion in each patient was delineated in T_(2)WI.Imaging features were extracted by the texture analysis software(Omni-Kinertics GE)that extracted parameters of the primary gray histogram and secondary texture features,including energy,entropy,skewness,kurtosis,uniformity,meanvalue and standard deviation.Statistical methods were used to compare the differences of MR-T_(2)WI texture parameters among prostate cancer,prostate hyperplasia nodules with T_(2)WI homogeneous low signal and prostate nodules with T_(2)WI heterogeneous low signal.Receiver operating curve(ROC)analysis and the area under the curve(AUC)were used for assessing the diagnostic efficiency of each parameter,and the differential diagnostic ability of quantitative texture parameters between the prostate cancer and prostate hyperplasia nodules with T_(2)WI low signal.Results In the quantitative parameters of MR-T_(2)WI texture analysis,there were statistical significant differences in energy,entropy,skewness,uniformity and standard deviation between groups of prostate cancer and prostate hyperplasia nodules,while kurtosis and meanvalue didn't show significant differences between both groups.For differentiating the prostate cancer from prostate hyperplasia nodules in standard deviation,the AUC was 0.832,which revealed the highest diagnosis efficiency,and the optimal cutoff point was 0.5804,with 68.75% sensitivity and 89.29% specificity.Moreover,entropy had both higher sensitivity and diagnostic efficiency.Conclusion Some quantitative parameters of MRT_(2)WI texture analysis were helpful for identifying the prostate cancer and benign prostate hyperplasia nodules with T_(2)WI low signal,providing the objective evidence for the differential diagnosis between them.