期刊文献+

纹理分析用于心肌淀粉样变性和肥厚型心肌病鉴别诊断的可行性研究 被引量:8

Feasibility study of myocardial texture analysis to differentiate between hypertrophic and amyloid cardiomyopathies
下载PDF
导出
摘要 目的:评价磁共振成像结合纹理分析技术对心肌淀粉样变性与肥厚型心肌病的鉴别诊断价值。方法:经心内膜活检证实的心肌淀粉样变性轻链型(CA)患者18例及经临床确诊的肥厚型心肌病患者18例,均采用3.0T MR标准化扫描获得电影序列图像,使用TexRad软件对所有患者左心室中段两腔短轴位图像进行取样分析,选择不同的空间缩放因子(SSF,取值为0和2~6),获得在不同纹理过滤尺度下代表纹理特征的5个参数值(包括均值、标准差、峰度、偏度、熵值),采用Mann-Whitney U检验进行组间比较,并使用ROC曲线评估各参数的诊断效能。结果:心肌淀粉样变性和肥厚型心肌病的熵值、峰度和偏度间的差异均有统计学意义(P<0.05)。诊断心肌淀粉样变性ROC曲线分析结果显示,SSF=3时取熵值≥2.68,诊断淀粉样变性的曲线下面积最大(0.87),诊断敏度87.2%,特异度为90.9%、符合率为88.0%。结论:纹理分析可用于心肌淀粉样变性和肥厚型心肌病的鉴别诊断,相比传统鉴别方法具有无需使用对比剂且缩短检查时间的优势。 Objective.. To investigate whether texture analysis can be used to differentiate hypertrophic cardiomyopa- thy (HCM) and cardiac amyloidosis (CA) on MR images. Methods.. Eighteen patients with CA and 18 patients with HCM were included in this study. Patients with CA were all confirmed by myocardial biopsy, and patients with HCM met the diagnostic criteria. All the patients had 3.0T magnetic resonance imaging before diagnosis. Texture analysis using a filtra- tion-histogram technique was performed using research software algorithm (TexRad) on end-diastolic frame of a mid short axis cine image. Five parameters (including mean, standard deviation, entropy, kurtosis and skewness) of texture feature were extracted using different value of spatial scale filter (SSF) between 0 and 6. Results:There were significant differences in entropy, skewness and kurtosis between HCM group and CA group. Receiver operation characteristic curve analysis showed that when entropy value of the septum segment of the LV wall 〉2. 68 at medium texture scale (SSF=3) ,there was the biggest AUC of 0. 87 for identification of CA from HCM (P〈0. 001), the sensitivity was 87. 2%, specitivity was 90.9 %, and accuracy was 88.0 %. Conclusion:TA of conventional cine images may complement conventional myocardial MR imaging to differentiate between hypertrophic and amyloid cardiomyopathies.
出处 《放射学实践》 北大核心 2017年第12期1225-1228,共4页 Radiologic Practice
基金 科技部"十三五"国家重点研发计划项目子课题(2016YFC1300402) 2017年度北京市科技新星计划交叉学科合作课题(Z171100001117136) 国家自然科学基金面上项目(81471725)
关键词 纹理分析 磁共振成像 淀粉样变性 肥厚型心肌病 Texture analysis Magnetic resonance imaging Amyloidosis Hypertrophic cardiomyopathy
  • 相关文献

同被引文献46

引证文献8

二级引证文献42

相关作者

内容加载中请稍等...

相关机构

内容加载中请稍等...

相关主题

内容加载中请稍等...

浏览历史

内容加载中请稍等...
;
使用帮助 返回顶部