Information-theoretic Limits, Algorithms, and Experiments to Validate Utility of High-density EEG in Clinical and Neuroscientific Settings
Electroencephalography (EEG), a noninvasive technology that records brain signals using electrodes placed on the scalp, is the most commonly used modality to diagnose many brain disorders. Despite EEG’s immense benefits of low-cost and portability, it is widely believed that EEG systems can only provide coarse, spatially low-resolution images of brain activity. Our recent theoretical and computational work [Grover & Venkatesh, Proc. IEEE ‘17] [Venkatesh & Grover, ISIT ‘17] [Grover et al. Allerton ’15], in part corroborated by ongoing experiments [Robinson et al., CNBC retreat ‘17] [Boring et al., ITA ‘16], challenges this belief about EEG’s low resolution. We demonstrate EEG’s true potential by (a) obtaining new fundamental limits that show that EEG can achieve far better image resolutions than what is currently assumed; (b) developing algorithms that dramatically outperform existing algorithms, and achieve the aforementioned fundamental limits; (c) validating our algorithm on experimental data to show its utility in real-world applications.