Mechanisms of high-resolution functional imaging and of decoding information conveyed by cortical columns

我的研究计划的长期目标是阐明功能磁共振成像的神经生理学和血液动力学机制。未来5年的目标是阐明高分辨率fMRI和基于fMRI的精细皮层结构中传递的信息解码的机制。 应用于人类功能磁共振成像数据的多变量机器学习算法可以解码由皮质列传达的信息，尽管体素大小相对于列的宽度很大。最初的推测表明，高空间频率分量的混叠可能是这一现象的基础。然而，我们已经证明这些猜测是错误的。还提出了若干其他机制。这些捐款来自：（I）局部不规则的柱状组织，（II）大规模的非柱状组织，（III）功能选择性静脉有偏见的引流区，和（IV）复杂的时空过滤的神经元活动的fMRI体素，可能携带信息的精细尺度的神经生理反应。尽管大量的人体成像研究使用基于精细结构中传达的信息的解码，但与潜在的神经活动和血液动力学机制的联系仍然很脆弱。 我们假设，成功解码的信息所传达的皮质柱依赖于上述机制（I）-（IV）的组合。 我们的第一个目的是确定在何种程度上不同的空间成分的血流动力学反应，包括灰质和宏观静脉的不同直径的反应，进行信息的神经元反应组织在皮质柱。 即使我们发现，粗糙的重新采样的空间模式的反应进行信息的精细尺度结构组织的神经元的反应，他们的贡献可能只占整体信息的一部分，可用于解码算法。因此，在我们的第二个目标中，我们将测试利用由时空响应而不是沿着沿着空间维度的响应所传达的信息是否单独地为成功解码贡献了附加信息。 在目标3中，我们将比较大规模的组织和眼优势柱结构的不规则性在多大程度上有助于成功解码。 在目标4中，我们将确定通过大规模组织或不规则性的贡献成功解码的程度取决于宏观静脉和小静脉的反应。 在我们的第五个目标中，我们将量化血氧和血容量响应相对于神经生理响应的时空点扩散函数。 我们的研究将促进高分辨率功能磁共振成像，以及解码精细结构传达的大脑反应的方法。

The long term goal of my research program is to elucidate the neurophysiological and hemodynamic mechanisms underlying fMRI. The goal for the upcoming 5 years is to elucidate the mechanisms underlying high-resolution fMRI and fMRI-based decoding of information conveyed in fine cortical structures. Multivariate machine learning algorithms applied to human fMRI data can decode information conveyed by cortical columns, despite the voxel size being large relative to the width of the columns. Initial speculations suggested that aliasing of high spatial-frequency components could underlie this phenomenon. However, we have proven that these speculations were wrong. Several other mechanisms have been proposed. These include contributions from: (I) local irregularities in the columnar organization, (II) large-scale non-columnar organizations, (III) functionally selective veins with biased draining regions, and (IV) complex spatiotemporal filtering of neuronal activity by fMRI voxels that may carry information on the fine-scale neurophysiological response. Although a large body of human imaging studies uses decoding based on information conveyed in fine structures, the link to the underlying neural activity and hemodynamic mechanisms remains tenuous. We hypothesize that the successful decoding of information conveyed by cortical columns relies on a combination of the mechanisms (I)-(IV) mentioned above.   Our first aim is to determine the extent to which different spatial components of the hemodynamic response, including the responses of gray matter and macroscopic veins of different diameters, carry information on neuronal responses organized in cortical columns. Even if we find that coarse resamples of the spatial patterns of the responses do carry information on the fine-scale structure organized neuronal responses, their contributions may account for only part of the overall information available to decoding algorithms. Therefore, in our second aim, we will test whether utilizing information conveyed by the spatiotemporal response rather than the response along the spatial dimension alone contributes additional information towards successful decoding. In aim 3, we will compare the extent to which large-scale organizations and irregularities in the structure of ocular dominance columns contribute to successful decoding. In aim 4, we will determine the extent to which successful decoding through contributions from large-scale organizations or irregularities depends on responses from macroscopic veins and venules. In our fifth objective, we will quantify the spatiotemporal point-spread function of the blood-oxygenation and blood volume responses relative to the neurophysiological response.  Our studies will advances high-resolution functional MRI, and the method of decoding brain responses conveyed by fine structure.