There is a critical need to develop network analysis tools that can assess changes in effective connectivity, i.e., in the cause-and-effect interactions among cortical regions, to complement and augment network analyses based on correlations between neural activity in different areas, such as resting state fMRI studies. Transcranial magnetic stimulation compatible electroencephalography (TMS/EEG) has unique potential for identifying causal network models because it offers a controlled, repeatable means for exogenous perturbation of brain activity and noninvasive measurement of the resultant effects with high temporal resolution and high signal-to-noise ratio. The long-term goal of this research is to develop new signal procesing tols for noninvasive assessment of cause and effect interactions in the cortex. The objective of this project, which is directed toward achieving this long-term goal, is to develop and evaluate a novel quantitative method that integrates the modeling of distributed cortical sources, the effect of TMS perturbation, and causal dynamic linear network models. The folowing specific aims wil be pursued to achieve this objective: 1) Develop and validate linear dynamic perturbation models (LDPM) for TMS/EEG;and 2) Quantify LDPM network properties in humans during three vigilance states: wake, non-REM sleep, and anesthesia. The project is innovative both in terms of the technical approach and the application to the study of the neural substrates of vigilance states. First, the LDPM model simultaneously accounts for the spatio-temporal attributes of TMS stimulation and spontaneous activity. Second, advanced signal-processing methods and EEG source models are employed to infer cortical network models directly from scalp recordings within a state-space framework. Third, cross- validation strategies provide definitive, quantitative evidence of model effectiveness. Fourth, use of human data from distinct vigilance states will provide rich diversity for model validation and quantitative characterization of the distinct network properties of the brain in wake, sleep, and anesthesia. The LDPM method is also applicable to task-related studies of specific networks, e.g., working memory. The project is significant because the LDPM method for causal network modeling and analysis of TMS/EEG data will provide neuroscientists with a new tool for analyzing cause and effect interactions in the cortex. This methodology will enable hypotheses about normal brain function, development, and disease to be tested and have an important positive impact on the characterization/diagnosis/treatment of connectivity disorders.

Public Health Relevance

The perturb-and-measure approach proposed here for identifying causal brain network models is expected to result in new tools for assessing the network properties of the thalamocortical system and is uniquely suitable for uncooperative patients, such as those in coma, vegetative, or minimally conscious states, or those with locked-in syndrome. It is applicable to assessment of any disease that affects the interaction between distinct neural systems in the brain, including stroke, schizophrenia, tumors, Alzheimers and epilepsy.

National Institute of Health (NIH)
National Institute of Biomedical Imaging and Bioengineering (NIBIB)
Exploratory/Developmental Grants (R21)
Project #
Application #
Study Section
Special Emphasis Panel (NOIT)
Program Officer
Sastre, Antonio
Project Start
Project End
Budget Start
Budget End
Support Year
Fiscal Year
Total Cost
Indirect Cost
University of Wisconsin Madison
Engineering (All Types)
Schools of Engineering
United States
Zip Code
Dentico, Daniela; Cheung, Bing Leung; Chang, Jui-Yang et al. (2014) Reversal of cortical information flow during visual imagery as compared to visual perception. Neuroimage 100:237-43