论文部分内容阅读
目的:比较目测法与外标记法对SPECT/CT图像进行配准的准确性;优化外标记法配准时标记点的个数及分布。方法:利用二维运动平台系统模拟人的呼吸运动,用配准后N个外标记点在两种图像间位置平均偏差(f)作为评价指标来衡量两图像配准的精度。结果:体模静止时,不同配准方法的f值间差别不大。而在不同运动状态下对7种方法配准得出的f值进行多组间的单因素方差分析,结果表明,用≥5个标记点配准的方法均与目测法、3点配准法间差异有统计学意义,P<0.05,且前者优于后者,而用≥5个标记点配准方法差异无统计学意义,P>0.05;不同研究者间目测法得到的f值差别较大;对5点配准法进行优化表明,标记点非共面分布优于标记点共面分布。结论:目测法配准受研究者主观因素的影响较大,准确性无法保证;用非共面5个外标记点对SPECT/CT图像进行融合配准是一种准确且简单易行的方法。
OBJECTIVE: To compare the accuracy of the SPECT / CT image registration between the visual method and the external marking method, and to optimize the number and distribution of the marking points of the external marking method. Methods: The two-dimensional motion platform system was used to simulate human respiratory motion. The registration accuracy of the two images was measured by using the average position deviation (f) between the two outer N points of the two registration images as the evaluation index. Results: When the phantom is stationary, the difference between f values of different registration methods is not significant. The results of single factor analysis of variance (ANOVA) between multiple groups of f values obtained from the registration of seven methods under different exercise states showed that the methods of registration with ≥5 markers were all consistent with the visual method, 3-point registration P <0.05, and the former is better than the latter. However, there is no significant difference between the two methods (P> 0.05). The differences of f values between different researchers The optimization of the 5-point registration method shows that the non-coplanar distributions of the markers are better than the coplanar ones. Conclusion: The visual inspection registration is greatly influenced by the subjective factors of the investigators, and its accuracy can not be guaranteed. It is an accurate and simple method to register the SPECT / CT images with non-coplanar five external markers.