High Spatio-temporal Resolution Noninvasive Neuroimaging using Diffuse Optical Tomography and Electroencephalography

Neuroimaging is of great importance in both clinical and neuroscientific applications such as disease diagnosis, functional mapping of the brain, etc., and for many of them, e.g. localization of the epileptic foci, requires both spatial and temporal resolution to be high. While it is desirable to image the brain noninvasively using techniques such as electroencephalography (EEG) and diffuse optics, they usually have limited resolution, either spatially or temporally, making them insufficient for such purposes when used alone.

The aim of this thesis is to show the possibility of noninvasive measurement of neuronal activity with high spatio-temporal resolution by combining the information of complementary modalities. In this thesis, two modalities, namely EEG and diffuse optical tomography (DOT), are considered, with the former measuring the electrical activity and the latter measuring the hemodynamic activity of the brain. Based on the assumption of intact neurovascular coupling, or specifically, that neuronal activity and hemodynamic changes are closely coupled both spatially and temporally, I show that the high temporal resolution of EEG and the high spatial resolution (that is, in comparison to EEG) of DOT can be combined to obtain a high spatio-temporal reconstruction of the brain activity.

The proposed method relies on two important assumptions: experimentally, sufficient spatial sampling of the brain through scalp measurements, and theoretically, a healthy neurovascular coupling assumption. To ensure that the first assumption holds, the thesis proposes an algorithm for optimized placement of EEG electrodes and DOT optodes given a certain region of interest, which can also be applied to many general sensor selection problems of underdetermined systems. To study the impact of pathological neurovascular coupling on hemodynamics and the mechanisms therein, in this thesis, I also propose a mathematical model which covers the entire neurovascular signal chain from neurons to arterioles, and demonstrate its applicability using cortical spreading depolarization (CSD) as an example.

Overall, this thesis offers a novel method for high spatio-temporal resolution brain activity reconstruction by integrating the information from simultaneously recorded EEG and DOT when neurovascular coupling is intact, and also an algorithm for the optimized placement of optodes to ensure good spatial accuracy. Further, a mathematical model of neurovascular coupling is proposed to study the mechanisms and the impacts on the hemodynamic recordings in pathological conditions.