Tudela Raúl, Muñoz-Moreno Emma, López-Gil Xavier, Soria Guadalupe
CIBER de Bioingeniería, Biomateriales y Nanomedicina (CIBER-BBN), Barcelona, Spain.
Experimental MRI 7T Unit, IDIBAPS, Barcelona, Spain.
PLoS One. 2017 Jan 24;12(1):e0170703. doi: 10.1371/journal.pone.0170703. eCollection 2017.
Diffusion-weighted imaging (DWI) quantifies water molecule diffusion within tissues and is becoming an increasingly used technique. However, it is very challenging as correct quantification depends on many different factors, ranging from acquisition parameters to a long pipeline of image processing. In this work, we investigated the influence of voxel geometry on diffusion analysis, comparing different acquisition orientations as well as isometric and anisometric voxels. Diffusion-weighted images of one rat brain were acquired with four different voxel geometries (one isometric and three anisometric in different directions) and three different encoding orientations (coronal, axial and sagittal). Diffusion tensor scalar measurements, tractography and the brain structural connectome were analyzed for each of the 12 acquisitions. The acquisition direction with respect to the main magnetic field orientation affected the diffusion results. When the acquisition slice-encoding direction was not aligned with the main magnetic field, there were more artifacts and a lower signal-to-noise ratio that led to less anisotropic tensors (lower fractional anisotropic values), producing poorer quality results. The use of anisometric voxels generated statistically significant differences in the values of diffusion metrics in specific regions. It also elicited differences in tract reconstruction and in different graph metric values describing the brain networks. Our results highlight the importance of taking into account the geometric aspects of acquisitions, especially when comparing diffusion data acquired using different geometries.
扩散加权成像（DWI）可对组织内水分子扩散进行量化，且正成为一种越来越常用的技术。然而，这极具挑战性，因为正确的量化取决于许多不同因素，从采集参数到一长串图像处理流程。在这项工作中，我们研究了体素几何形状对扩散分析的影响，比较了不同的采集方向以及等轴和非等轴体素。用四种不同的体素几何形状（一种等轴和三种不同方向的非等轴）以及三种不同的编码方向（冠状面、轴位面和矢状面）采集了一只大鼠大脑的扩散加权图像。对这12次采集中的每一次都分析了扩散张量标量测量、纤维束成像和脑结构连接组。相对于主磁场方向的采集方向会影响扩散结果。当采集切片编码方向与主磁场不对齐时，会出现更多伪影且信噪比更低，这会导致各向异性张量减少（分数各向异性值更低），产生质量较差的结果。使用非等轴体素在特定区域的扩散指标值上产生了统计学上的显著差异。它还在纤维束重建以及描述脑网络的不同图指标值上引发了差异。我们的结果突出了考虑采集几何方面的重要性，尤其是在比较使用不同几何形状采集的扩散数据时。
