Department of Medical Biophysics, University of Toronto, Toronto, Ontario M5G 2M9, Canada.
Med Phys. 2011 May;38(5):2742-53. doi: 10.1118/1.3578600.
Tumor characterization employing a voxel-wise analysis of image signal facilitates the determination of the spatial distribution of tumor attributes, and when employed for therapy response assessment offers the promise of greater sensitivity to change than conventional approaches. However, the accuracy of a voxel-wise analysis of change is limited by local registration uncertainties that can disrupt the spatiotemporal correspondence between assessed voxels. We present a method for assessing voxel correspondence strength using a multiresolution local histogram-based measure of image structure similarity. When employed in a longitudinal tumor imaging context, a voxel similarity measure must be robust to intensity variations that can arise from the image acquisition, treatment effects, or changes in underlying disease processes. Consequently, the local histogram-based similarity measure is evaluated for sensitivity to structural change and robustness to intensity variation and is compared against normalized mutual information and normalized cross-correlation.
T1-weighted (T1W) magnetic resonance (MR) images of glioblastoma acquired as part of a longitudinal response assessment study are first rigidly registered, and then similarity between spatially corresponding voxels is evaluated using multiresolution local histograms. Region-based and nonuniform intensity changes of varying magnitude as well as deformations to image structure are applied individually and in combination to the test images. Statistical analysis is used to test for interaction effects between the applied perturbations and the value of the local histogram similarity function. Pair-wise voxel similarity maps are computed for pairs of longitudinal clinical MR image volumes and compared with observed patterns of change on conventional imaging.
The simulations demonstrated that the local histogram measure was robust to intensity modulations applied to increasing region sizes and exhibited a strong negative correlation with the magnitude of local deformation. No statistically significant interaction effects were observed upon the value of the local histogram similarity function when deformation was applied in conjunction with a nonuniform intensity change. Pair-wise voxel similarity maps were consistent with image change observed on T1W MR imaging and revealed patterns of change not apparent in conventional image sequences.
A measure of local histogram image structure similarity can be used to assess the strength of voxel to voxel correspondences independently of intensity nonuniformities. The metric can provide a local estimate of the limits of achievable correspondence underlying the registration and voxel-wise comparison of signal in longitudinal imaging used for assessing tumor response to treatment.
通过对图像信号进行体素分析来进行肿瘤特征描述,有助于确定肿瘤属性的空间分布,并且当用于治疗反应评估时,比传统方法具有更高的变化敏感性。然而,体素分析的准确性受到局部配准不确定性的限制,这种不确定性会破坏评估体素之间的时空对应关系。我们提出了一种使用基于多分辨率局部直方图的图像结构相似性度量来评估体素对应强度的方法。当应用于纵向肿瘤成像时,体素相似性度量必须对可能由于图像采集、治疗效果或潜在疾病过程变化引起的强度变化具有鲁棒性。因此,评估了基于局部直方图的相似性度量对结构变化的敏感性和对强度变化的鲁棒性,并与归一化互信息和归一化互相关进行了比较。
对胶质母细胞瘤进行的纵向反应评估研究中获取的 T1 加权(T1W)磁共振(MR)图像首先进行刚性配准,然后使用多分辨率局部直方图评估空间对应体素之间的相似性。分别和组合地将不同幅度的区域和非均匀强度变化以及图像结构的变形应用于测试图像。统计分析用于测试施加的扰动与局部直方图相似性函数值之间的相互作用效应。计算了纵向临床 MR 图像体积对的成对体素相似性图,并与常规成像上观察到的变化模式进行了比较。
模拟结果表明,局部直方图测度对施加到增大区域大小的强度调制具有鲁棒性,并且与局部变形的幅度呈强烈负相关。当变形与非均匀强度变化结合使用时,在局部直方图相似性函数值上未观察到统计学上显著的相互作用效应。成对体素相似性图与 T1W 磁共振成像上观察到的图像变化一致,并揭示了常规图像序列中不明显的变化模式。
可以使用局部直方图图像结构相似性度量来评估体素之间对应关系的强度,而与强度不均匀性无关。该度量可以提供注册和用于评估治疗后肿瘤反应的信号体素分析的基础上,实现对应关系的局部估计。