Tools for Large-Scale Platform-Independent MEG Data Analysis

点击翻译按钮获取中文摘要

DESCRIPTION (provided by applicant): Magnetoencephalography (MEG) provides a unique window on the large scale spatiotemporal neural processes that underlie human brain function. However, even with restrictive models, the low SNR environment, ill-posedness of the inverse problem, and difficulty of differentiating ongoing brain activity and other electrophysiological signals from induced and event-related changes result in unique challenges in data analysis and interpretation. Results obtained are dependent on factors such as the choice of the constraints imposed in the source estimation procedures, the forward model, the time windows chosen for the analyses, and preprocessing algorithms, so that the absence of standardization in the MEG community results in substantial differences in the analysis and findings from similar experiments conducted at different sites. This project will develop a standard work flow for the analysis of MEG data, coupled with specifications of data structures to enable cross site data sharing and comparison of results. To achieve this goal we will modify and extend two existing MEG software packages that have been developed independently: MNE at MGH and BrainStorm at USC. We will pursue four specific aims: (1) We will establish a unified workflow, data structures, and file formats for BrainStorm and MNE to enable sharing of data at all levels of analysis between the two packages. The workflow will also facilitate present and future applications requiring the fusion of data from imaging modalities other than MEG and anatomical MRI. (2) Using available scripting languages, we will develop an efficient mechanism for analyzing multiple data sets with an emphasis on computing individual and group statistics with thresholding to correct for multiple comparisons. These tools will be compatible with all major operating systems. (3) Using the MIND MEG Consortium human calibration study data we will establish standard procedures for verification and validation of the implementations of MEG inverse procedures. (4) We will distribute open source software through linked sites at MGH and USC, with associated documentation, discussion forums, and online tutorials. PUBLIC HEALTH RELEVANCE: Magnetoencephalography (MEG) is a totally non-invasive brain imaging tool which provides information on the spatial distribution and precise temporal orchestration of human brain activity. MEG can be thus used to understand and diagnose abnormalities underlying a wide range neurological and psychiatric illnesses including as epilepsy, schizophrenia, obsessive- compulsive disorder, autism spectrum disorders, and Alzheimer's disease, as well as cognitive deficits such as delayed acquisition of language. However, more widespread use of MEG especially in large populations has been problematic because of the lack of well-established analysis appproaches. This research will provide well-documented and tested analysis tools to promote both basic neuroscience and clinical research applications using MEG in combination with anatomical MRI.