Affine transformation registers small scale lung deformation.

Affine transformation registers small scale lung deformation.
复制标题

仿射变换记录小范围的肺变形。

DOI:
10.1109/embc.2012.6347190
复制
发表时间:
2012
期刊:
Annual International Conference of the IEEE Engineering in Medicine and Biology Society. IEEE Engineering in Medicine and Biology Society. Annual International Conference
影响因子:
--
通讯作者:
Theilmann,RebeccaJ
Theilmann,RebeccaJ
中科院分区:
--
文献类型:
--
作者:
Arai,TatsuyaJ;Villongco,ChristopherT;Villongco,MichaelT;Hopkins,SusanR;Theilmann,RebeccaJ

文献摘要

相似文献

为了评估多个肺磁共振图像之间的小尺度肺变形的性质,测试了两种不同的基于运动强度的图像配准技术:仿射和双三次Hermite插值。仿射方法估计在整个肺中均匀分布的变形度量。双三次Hermite方法允许表达不均匀分布的变形度量,如拉格朗日应变。使用心脏触发反转恢复技术,在FRC下获得4名健康受试者(年龄:28.5(6.2))右肺矢状面中肺血管结构的10个连续图像。一幅图像用作参考图像,其余图像(目标图像)使用两种图像配准技术扭曲到参考图像上。将肺域内参考图像与变换后的目标图像之间的归一化相关性作为优化的代价函数,并将图像强度差的均方根(RMS)用于评价配准质量。与未配准的靶图像相比,两种图像配准技术均显著改善了RMS(p= 0.04)。拉格朗日应变的空间平均值(µE)和标准差(σE)是根据双三次Hermite方法近似的肺变形的空间分布计算的,测量值为10− 3或更小,几乎可以忽略不计。结果,FRC肺体积之间的小尺度肺变形在空间上是均匀的,并且可以简单地通过仿射变形来表征,即使双三次Hermite方法能够表达肺变形的复杂空间图案。
To evaluate the nature of small scale lung deformation between multiple pulmonary magnetic resonance images, two different kinematic intensity based image registration techniques: affine and bicubic Hermite interpolation were tested. The affine method estimates uniformly distributed deformation metrics throughout the lung. The bicubic Hermite method allows the expression of heterogeneously distributed deformation metrics such as Lagrangian strain. A cardiac triggered inversion recovery technique was used to obtain 10 sequential images of pulmonary vessel structure in a sagittal plane in the right lung at FRC in 4 healthy subjects (Age: 28.5(6.2)). One image was used as the reference image, and the remaining images (target images) were warped onto the reference image using both image registration techniques. The normalized correlation between the reference and the transformed target images within the lung domain was used as a cost function for optimization, and the root mean square (RMS) of image intensity difference was used to evaluate the quality of the registration. Both image registration techniques significantly improved the RMS compared with non-registered target images (p= 0.04). The spatial mean (µE) and standard deviation (σE) of Lagrangian strain were computed based on the spatial distribution of lung deformation approximated by the bicubic Hermite method, and were measured on the order of 10−3or less, which is virtually negligible. As a result, small scale lung deformation between FRC lung volumes is spatially uniform, and can be simply characterized by affine deformation even though the bicubic Hermite method is capable of expressing complicated spatial patterns of lung deformation.