Neurovascular coupling links brain activity to local changes in blood flow, forming the basis for non-invasive brain mapping. Using multiscale imaging, we investigated how vascular activity spatially relates to neuronal activity elicited by single whiskers across different columns and layers of mouse cortex. Here we show that mesoscopic hemodynamic signals quantitatively reflect neuronal activity across space but are composed of a highly heterogeneous pattern of responses across individual vessel segments that is poorly predicted by local neuronal activity. Rather, this heterogeneity is dependent on vessel directionality, specifically in thalamocortical input layer 4, where capillaries respond preferentially to neuronal activity patterns along their downstream perfusion domain. Thus, capillaries fine-tune blood flow based on distant activity and encode laminar-specific activity patterns. These findings imply that vascular anatomy sets a resolution limit on functional imaging signals, where individual blood vessels inaccurately report neuronal activity in their immediate vicinity but, instead, integrate activity patterns along the vascular arbor.

神经血管耦合将大脑活动与血流的局部变化联系起来，构成了非侵入性脑映射的基础。使用多尺度成像，我们研究了血管活动在空间上如何与小鼠皮层不同柱和层的单须引起的神经元活动相关。在这里，我们表明，介观血流动力学信号定量地反映了跨空间的神经元活动，但由跨单个血管段的高度异质反应模式组成，局部神经元活动很难预测。相反，这种异质性取决于血管方向性，特别是在丘脑皮质输入层 4 中，其中毛细血管优先响应沿其下游灌注域的神经元活动模式。因此，毛细血管根据远处活动微调血流，并编码层状特异性活动模式。这些发现意味着血管解剖学为功能成像信号设定了分辨率限制，其中单个血管不准确地报告了它们附近的神经元活动，而是沿着血管乔木整合了活动模式。