摘要
目的探讨CT图像纹理分析在鉴别CT首诊发现的骨内硬化结节良恶性的潜在价值。方法搜集110例CT首诊发现骨内高密度病灶患者,均经过临床及影像资料随诊最终确定诊断。其中30例恶性肿瘤患者,78个成骨性转移病灶;80例非肿瘤患者,80个骨岛。经手动分割勾画病灶感兴趣区(ROI),应用MaZda软件分析病灶纹理特征参数。组间纹理参数比较采用独立样本t检验,并采用因子分析对纹理指标进行降维归类,采用受试者工作特征曲线(ROC)分析纹理指标鉴别良恶性结节的准确率、灵敏度、特异度。结果成骨性转移和骨岛组间灰度均值(154.9±16.3、207.5±29.7)、方差(402.9±238.7、754.5±462.5)、偏度(-0.3±0.5、-0.9±1.0)、平方和(24.6±15.1、39.7±26.4)等多个纹理指标具有显著差异(P<0.01)。因子分析得到三类相似性纹理指标即纹理明暗度相关指标、纹理复杂度相关指标、纹理周期性相关指标,ROC分析图像明暗度相关纹理指标诊断准确率较高,曲线下面积0.99,灵敏度、特异度分别为91%、88%。结论CT纹理学分析方法能够为成骨性转移和骨岛的鉴别提供帮助。
Objective To investigate the potential value of CT image texture analysis in identifying the benign and ma-lignant intracranial sclerosis found in CT. Methods A total of 110 patients with high-intensity lesions found in the firstCT examination were diagnosed by clinical and imaging data. Among them,30 patients with malignant tumors,78 with os-teogenic metastases; 80 with non-tumor patients and 80 with bone islands. The lesion region of interest ( ROI) was manual-ly segmented,and the MaZda software was used to analyze the lesion texture parameters. The comparison of the texture pa-rameters between groups was performed by independent sample t test,and the texture index was classified by dimensional a-nalysis. The receiver operating characteristic curve ( ROC) was used to analyze the texture index to identify the accuracy,sensitivity and specificity of benign and malignant nodules. Results The mean gray scale between the osteogenic metasta-sis and the bone island group ( 154. 9 ±16. 3,207. 5 ±29. 7) ,variance ( 402. 9 ±238. 7,754. 5 ±462. 5) ,skewness ( -0. 3 ±0. 5,-0. 9 ±1. 0) ,square sum ( 24. 6 ±15. 1,39. 7 ±26. 4) and other texture indicators have significant differ-ences ( P <0. 01) . Factor analysis results in three types of similarity texture indicators,namely,texture-darkness-relatedindicators,texture-complexity-related indicators,and texture-periodic-related indicators. ROC analysis image shading-relat-ed texture indicators have higher diagnostic accuracy,and the area under the curve is 0. 99. Sensitivity and specificity. Thedegrees are 91% and 88% respectively. Conclusion CT texture analysis can help to differentiate osseous metastasis andbone island.
作者
马军超
卢彬平
许晓通
吴虎林
王瑞峰
MA Junchao;LU Binping;XU Xiaotong(Department of Radiology,Affiliated Hospital of Shaanxi University of Chinese Medicine,Xianyang 712000,P. R. China)
出处
《临床放射学杂志》
CSCD
北大核心
2018年第11期1897-1900,共4页
Journal of Clinical Radiology
关键词
纹理特征
骨岛
成骨性转移
Texture features
Bone Island
Osteoblastic metastases
