Evaluation of 3D modality-independent elastography for breast imaging: a simulation study
Evaluation of 3D modality-independent elastography for breast imaging: a simulation study
复制标题
DOI：
10.1088/0031-9155/53/1/010
复制
发表时间：
2008-01-07
影响因子：
3.5
通讯作者：
Miga, M. I.
中科院分区：
文献类型：
--
作者：
Ou, J. J.;Ong, R. E.;Miga, M. I.
本文报告的开发和初步测试的三维实施的反问题技术提取软组织弹性信息，通过非刚性模型为基础的图像配准。模态无关弹性成像（MIE）算法调整生物力学模型的弹性特性，以实现在不同静态载荷状态下采集的图像之间的最大相似性。一系列的模拟实验与人类乳房的临床图像集进行了测试的能力的方法来识别和表征一个放射摄影隐匿性僵硬病变。由于边界条件是算法的关键输入，因此在系统和随机噪声过程的背景下比较了三种半自动表面点对应方法。结果表明，3D MIE能够成功地重建弹性图像使用从磁共振和X射线计算机断层扫描系统获得的数据。在所有病例中，病变均被正确定位，其相对弹性合理地接近真实值（使用空间先验的为3.5%，未使用空间先验的为11.6%）。此外，薄板样条插值的表面配准的不准确性没有超过不可接受的边界条件误差的经验阈值。
This paper reports on the development and preliminary testing of a three-dimensional implementation of an inverse problem technique for extracting soft-tissue elasticity information via non-rigid model-based image registration. The modality-independent elastography (MIE) algorithm adjusts the elastic properties of a biomechanical model to achieve maximal similarity between images acquired under different states of static loading. A series of simulation experiments with clinical image sets of human breasts were performed to test the ability of the method to identify and characterize a radiographically occult stiff lesion. Because boundary conditions are a critical input to the algorithm, a comparison of three methods for semi-automated surface point correspondence was conducted in the context of systematic and randomized noise processes. The results illustrate that 3D MIE was able to successfully reconstruct elasticity images using data obtained from both magnetic resonance and x-ray computed tomography systems. The lesion was localized correctly in all cases and its relative elasticity found to be reasonably close to the true values (3.5% with the use of spatial priors and 11.6% without). In addition, the inaccuracies of surface registration performed with thin-plate spline interpolation did not exceed empiric thresholds of unacceptable boundary condition error.