Athinoula A. Martinos Center for Biomedical Imaging, Department of Radiology, Harvard Medical School, Massachusetts General Hospital, 149 13th St., Charlestown, Boston, MA 02129, USA.
Max Planck Institute for Biological Cybernetics, Max-Planck-Ring 11, 72076 Tübingen, Germany.
Neuroimage. 2019 Aug 1;196:337-350. doi: 10.1016/j.neuroimage.2019.04.036. Epub 2019 Apr 17.
Functional magnetic resonance imaging (fMRI) is now capable of sub-millimetre scale measurements over the entire human brain, however with such high resolutions each voxel is influenced by the local fine-scale details of the cerebral cortical vascular anatomy. The cortical vasculature is structured with the pial vessels lying tangentially along the grey matter surface, intracortical diving arterioles and ascending venules running perpendicularly to the surface, and a randomly oriented capillary network within the parenchyma. It is well-known that the amplitude of the blood-oxygenation level dependent (BOLD) signal emanating from a vessel depends on its orientation relative to the B-field. Thus the vascular geometric hierarchy will impart an orientation dependence to the BOLD signal amplitudes and amplitude differences due to orientation differences constitute a bias for interpreting neuronal activity. Here, we demonstrate a clear effect of cortical orientation to B in the resting-state BOLD-fMRI amplitude (quantified as the coefficient of temporal signal variation) for 1.1 mm isotropic data at 7T and 2 mm isotropic at 3T. The maximum bias, i.e. the fluctuation amplitude difference between regions where cortex is perpendicular to vs. parallel to B, is about +70% at the pial surface at 7T and +11% at 3T. The B orientation bias declines with cortical depth, becomes progressively smaller closer to the white matter surface, but then increases again to a local maximum within the white matter just beneath the cortical grey matter, suggesting a distinct tangential network of white matter vessels that also generate a BOLD orientation effect. We further found significant (negative) biases with the cortex orientation to the anterior-posterior anatomical axis of the head: a maximum negative bias of about -30% at the pial surface at 7T and about -13% at 3T. The amount of signal variance explained by the low frequency drift, motion and the respiratory cycle also showed a cortical orientation dependence; only the cardiac cycle induced signal variance was independent of cortical orientation, suggesting that the cardiac induced component of the image time-series fluctuations is not related to a significant change in susceptibility. Although these orientation effects represent a signal bias, and are likely to be a nuisance in high-resolution analyses, they may help characterize the vascular influences on candidate fMRI acquisitions and, thereby, may be exploited to improve the neuronal specificity of fMRI.
功能磁共振成像（fMRI）现在能够在整个大脑中进行亚毫米级的测量，然而，由于如此高的分辨率，每个体素都会受到大脑皮质血管解剖学的局部细微细节的影响。皮质血管结构具有沿灰质表面切线分布的软脑膜血管，穿过皮质的直下动脉和向心静脉垂直于表面，以及在实质内随机定向的毛细血管网络。众所周知，来自血管的血氧水平依赖（BOLD）信号的幅度取决于其相对于 B 场的方向。因此，血管几何层次结构将赋予 BOLD 信号幅度以方向依赖性，并且由于方向差异而导致的幅度差异构成了解释神经元活动的偏差。在这里，我们在 7T 时的 1.1mm 各向同性数据和 3T 时的 2mm 各向同性数据中，证明了皮质方向对静息状态 BOLD-fMRI 幅度（以时间信号变化系数量化）的明显影响。最大偏差，即在皮层垂直于 B 与平行于 B 的区域之间的波动幅度差异，在 7T 时约为软脑膜表面处的+70%，在 3T 时约为+11%。B 方向偏差随皮质深度而降低，在靠近白质表面处逐渐变小，但随后在白质内再次增加到皮质灰质下方的局部最大值，表明白质中有一个明显的切线血管网络也会产生 BOLD 方向效应。我们还发现，皮质相对于头部的前后解剖轴的方向存在显著的（负）偏差：在 7T 时软脑膜表面处最大负偏差约为-30%，在 3T 时约为-13%。低频漂移、运动和呼吸周期解释的信号方差也表现出皮质方向依赖性；只有心脏周期引起的信号方差与皮质方向无关，这表明图像时间序列波动的心脏诱导分量与磁化率的显著变化无关。尽管这些方向效应代表了信号偏差，并且可能在高分辨率分析中造成干扰，但它们可能有助于描述血管对候选 fMRI 采集的影响，从而可以利用它们来提高 fMRI 的神经元特异性。
