Ph.D. Thesis Defense - M. Tanveer Talukdar

Dartmouth Events

Ph.D. Thesis Defense - M. Tanveer Talukdar

“Modeling Neurovascular Coupling with Applications in Multimodal EEG-NIRS”

Thursday, July 31, 2014
Jackson Conf Room, Cummings Hall
Intended Audience(s): Public


Thesis Committee

Solomon G. Diamond, Ph.D.  (Chair)

Erik J. Kobylarz, Ph.D.

Minh Q. Phan, Ph.D.

Dana H. Brooks, Ph.D.




Neurovascular coupling is a term used to described changes in local blood flow in the brain during neural activity.  This is a complex biophysical process that is critical for normal functioning of the brain. Surprisingly little is understood about the mechanistic basis of this neurovascular coupling relationship.  More importantly, alternation in the neurovascular coupling relationship has been implicated in neurodegenerative disorders like stroke, hypertension, epileptic seizures and Alzheimer's. In this work, we investigate neurovascular coupling in the human brain using a non-invasive multimodal neuroimaging technique. We used simultaneous electroencephalography (EEG) and near-infrared spectroscopy (NIRS) to measure the electro-cortical and hemodynamic response during activation of the somatosensory cortex. We then developed a linear parametric model that uses gamma transfer functions to relate EEG spectral envelopes to the changes in NIRS hemodynamics. The EEG spectral envelopes reflect time-varying power variations in neural rhythms that have temporal scales close to the hemodynamic response. This makes it practical to use gamma transfer functions, which can vary in their shape and peak times, to predict NIRS hemodynamics as output from EEG spectral envelopes as input.  Based on simulated and experimental EEG-NIRS data obtained from human subjects, we show that multiple gamma transfer functions can be used to describe the neurovascular coupling relationship. Also, after clustering the gamma transfer function parameters recovered from fitting experimental EEG-NIRS data, statistically significant parameters sets were obtained for each subject (two sided Z-test, p < 0.05). These clustered parameters sets were next used to predict NIRS hemodynamics from EEG spectral envelopes. The correlations between the predicted hemodynamics and measured NIRS hemodynamics were then mapped onto a network showing the connectivity between the electro-cortical response and changes in hemodynamics during neural activity. Network analysis revealed significant differences in the number of links between the left and right hemisphere (two sided Z-test, N = 5, p < 0.05). These neurovascular-coupling networks could have important implication for clinical neuroimaging studies.

For more information, contact:
Daryl Laware

Events are free and open to the public unless otherwise noted.