Independent Component Analysis Optimized for fMRI

点击翻译按钮获取中文摘要

This application requests support to continue our work in developing robust, independent component analysis (ICA) methods for measuring brain activity using functional magnetic Resonace Imaging (fMRI), and applying these methods to a population of brain tumor patients. The development of an accurate, noninvasive method to locate eloquent cortex in relation to brain pathologies is a goal of critical importance to the neurosurgical community. Conventional linear regression analysis methods, while robust in healthy, motivated volunteers, often fail in clinical situations. Clinical cases are frequently marred by patient movement, and failure of the patient to correctly perform the task. Our goal behind this project is to use the knowledge and experience we have gained thus far to develop a more robust ICA-based method that takes into account all of the relevant physiological parameters that describe the fMRI signal and allows a statistical significance level to to be attached to the ICA results. The primary aims of the project are: 1. Optimize the ICA algorithm and paradigms for fMRI. 2. Compare the results of an ICA and a standard regression analysis in healthy volunteers. 3. Develop a frequency representation of the fMRI time course as a means to facilitate identification of task related components in an ICA. 4. Develop a hybrid ICA method whereby time courses are computed with an ICA of a fMRI data set to build a linear model of the fMRI data set in the frequency domain.