Multimodal integration of functional neuroimaging data

现在越来越多的非侵入性神经成像技术可用于研究正常的大脑功能，也可用于研究大脑疾病。然而，目前还不存在任何理想的探测技术，能够精确地测量大脑活动发生的“时间”（以毫秒计）和“地点”（以毫米计）。每种神经成像技术都是通过不同的机制（神经元电、氧气消耗）来测量大脑活动的一部分，有时更专注于“在哪里？”（良好的空间分辨率）或者“什么时候？”（时间分辨率好）。每种技术都有其特定的优点和局限性，单独分析每种技术只能提供潜在大脑活动的部分信息。我的研究项目的目标是开发适当组合这些多模态数据的方法，即数据融合方法，以便通过单独考虑每种模态来检测可能遗漏的额外信息。一个典型的挑战是将高时间分辨率直接测量神经元活动的模式与高空间分辨率间接测量相同功能的其他模式结合起来，例如通过血液动力学过程。

More and more non-invasive neuroimaging techniques are now available to study normal brain functions but also brain disorders. However, it does not exist any ideal exploration technique able to measure precisely "when" (at a ms scale) and "where" (at a mm scale) an activity occurred in the brain. Each neuroimaging technique measures through different mechanisms (neuronal electricity, oxygen consumption) part of brain activity, being sometimes more dedicated to the question "where?" (good spatial resolution) or to the question "when?" (good temporal resolution). Each of these techniques having its specific advantages and limitations, analyzing each of them individually only provides partial information on underlying brain activity. The objective of my research project is to develop methods to appropriately combine these multimodal data, i.e., data fusion methods, in order to detect additional information that could be missed by considering each modality individually. A typical challenge is to combine modalities measuring directly neuronal activity with high temporal resolution with other modalities measuring indirectly the same function with high spatial resolution, through hemodynamic processes for instance.