自动分割方法分析心脏MRI左心室功能的可行性

摘要

Objective:To investigate the feasibility of automatic segmentation method on left ventricle function analysis of real-time cardiac MRI. Methods:Cardiac magnetic resonance data of 12 volun-teers were collected. The automatic segmentation methods of subtraction image,Hough transformation and K-mean clustering were used to segment left ventricle;two observers( observer 1 and observer 2 ) with more than one year heart disease data analysis experience to gain EDV,ESV,sSV and EF by means of automatic segmentation and manual segmentation,then evaluating the feasibility of automatic segmentation method on left ventricle function analysis of real-time cardiac MRI. Results:The result shows that EDV without statistical significant difference,ESV,SV and EF data collected manually or automatically with statistical significant difference( P ﹤0 . 05 ). EDV difference showed no statistical significance( P﹥0 . 05 ). Conclusion:Automatic segmentation method could analyze the EDV function of left ventricle.%目的：探讨自动分割方法对实时心脏磁共振成像（ MRI）左心室功能分析的可行性。方法：采集12名自愿者从心尖层到基底层实时MRI心脏成像数据，运用差异图像、Hough变换及K均值聚类的自动分割方法分割左心室，由两名具有1年以上心脏数据分析经验的观察者（观察者1和观察者2）运用自动分割和手动分割方法获取左心室舒张末期容量（ EDV）、收缩末期容量（ ESV）、每博血量（ SV）和射血分数（ EF）等功能参数，评价心脏实时MRI自动分割方法分析左心室功能的可行性。结果：自动分割法可以对舒张末期和收缩末期图像进行分割，但与手动分割方法相比，实际分割曲线仍存在部分差异；两位观察者手动或自动分割法采集的ESV、SV及EF数据差异有统计学意义（P﹤0.05），EDV差异无统计学意义（P﹥0.05）。结论：心脏实时MRI自动分割方法能够对左心室EDV功能参数进行分析。