Skip to main content
eScholarship
Open Access Publications from the University of California

UCLA

UCLA Electronic Theses and Dissertations bannerUCLA

Neural Signal Processing: Electrode-based Brain Imaging, Focalized Neural Stimulation, and Neural Dynamics Study

Abstract

Simultaneous neuroimaging and neurostimulation provides a powerful tool for monitoring the functional state of the nervous system as well as treating neural diseases. The neuroimaging is able to obtain real-time information of the targets, and provides a dynamic guidance for the neurostimulation, so that the underlying neural network can be modulated with high precision.

The first part of this thesis aims to develop EEG-based brain imaging algorithms with high reconstruction accuracy and speed. EEG brain imaging is able to produce brain images with excellent temporal resolution (~ms), and is therefore a good candidate for studying the dynamic brain states. However, the corresponding EEG inverse problem is highly ill-posed, thus requiring regularization techniques to impose additional constraints to obtain a precise result. We have developed two novel EEG-based brain imaging methods (s-SMOOTH and gFOTV) using sparse regularizations based on the compressed sensing principle - these methods demonstrate better performance than the state-of-the-art methods in terms of reconstruction accuracy, localization accuracy, and focalization degree. Furthermore, in order to obtain real-time brain images, a novel parallel computing algorithm has been developed to accelerate the image reconstruction speed.

The second part aims to develop optimization methods for noninvasive electrical stimulation, so as to provide high focal accuracy and desired intensity at the target under specific constraints. Conventional optimization methods either maximize the intensity at the target, resulting in low focal accuracy, or maximize the focal accuracy at the expense of low intensity. We have developed a novel optimization method called Stimulation with Optimal Focality and Intensity (SOFI), which provides both high intensity and focal accuracy within the safety constraints. We apply this method to transcranial current stimulation (tCS) and transcutaneous spinal cord stimulation (tSCS).

The last part further studies the neural dynamics with advanced time-frequency analysis techniques. We employ an accurate time-frequency analysis approach - Hilbert Huang Transform (HHT) - which is able to deal with nonstationary and nonlinear signals such as EEG/ECoG. We have demonstrated that it achieves better results than the widely used method - Fourier Transform (FT) - by comparing them in the applications of seizure detection and cross-frequency coupling.

Main Content
For improved accessibility of PDF content, download the file to your device.
Current View