Characterization of Visual Stimulus-Induced Gamma Oscillations as Signatures of Mechanisms Underlying Healthy Aging and Disease

Abstract

Neural oscillations are rhythmic fluctuations in the electrical activity recorded from the brain that convey important information about brain function. Gamma oscillations, which refer to the 30–80 Hz frequency range, are thought to reflect higher cognitive functions like attention, working memory, and perception. Recent studies have shown that stimulus-induced gamma recorded using electroencephalogram (EEG) in humans weakens with healthy aging and cognitive disorders. This thesis characterizes three aspects of these oscillations: reliability, functional connectivity (FC) and spatial extent. An EEG dataset comprising resting and stimulus-induced responses to large achromatic cartesian gratings over 247 subjects (227 healthy, 14 subjects with Mild Cognitive Impairment (MCI) and 6 subjects with early Alzheimer’s Disease (AD)) was considered for all the objectives addressed in the thesis.

The first aim focuses on the intra-subject reliability of alpha and gamma oscillations between two recordings from the same individual separated by 1 year or more. We found that alpha and gamma oscillations were highly reliable for a subject; measures such as change in band powers in alpha and gamma bands, temporal evolution of power within alpha and gamma bands, and the overall spectral profile succeeded in reliably identifying two recordings from the same subject with high accuracy, with an Area Under Curve (AUC) of 0.89 (Kumar et al., 2022; Cerebral Cortex Communications). These results establish the reliability of visual stimulus-induced features.

In the second aim, we found that stimulus-induced FC in the sensor space (computed from the data during the stimulus period) reduced with age in alpha (8‒12 Hz) and gamma (20‒34 Hz) even though power only reduced for gamma, suggesting that FC varies distinctly from power changes. Strikingly, the FC reduction with age persisted in alpha and gamma, even when changes in power were accounted for. Gamma FC weakened with cognitive disorders like MCI compared to the respective age and gender-matched control subjects, even when power differences were controlled for (Kumar and Ray, 2023, EJN).

For the final aim, we studied power in the source space using a technique called eLORETA (Exact low-resolution brain electromagnetic tomography analysis) within the alpha and gamma bands. The source distribution was parametrized as an exponential decay function with an amplitude and decay parameter in the inter-voxel separation space. This allowed us to test whether the change in source distribution can be explained by a decrease in magnitude with aging (change in amplitude parameter) or if further shrinkage (change in decay parameter) is also needed for explanation. Interestingly, the source distributions in the gamma band showed a reduction mainly in the amplitude of the source power with aging, while the alpha source power distribution remained indifferent. Altogether, the three studies characterize various features within the sensor and source space of stimulus-induced gamma oscillations, which could potentially be used as a biomarker for the early diagnosis of mental disorders such as MCI and AD.