Spatial information in large-scale neural recordings

The Listening Zone of Human Electrocorticographic Field Potential Recordings

Intracranial electroencephalographic (icEEG) recordings provide invaluable insights into neural dynamics in humans because of their unmatched spatiotemporal resolution. Yet, such recordings reflect the combined activity of multiple underlying generators, confounding the ability to resolve spatially distinct neural sources. To empirically quantify the listening zone of icEEG recordings, we computed correlations between signals as a function of distance (full width at half maximum; FWHM) between 8752 recording sites in 71 patients (33 female) implanted with either subdural electrodes (SDEs), stereo-encephalography electrodes (sEEG), or high-density sEEG electrodes. As expected, for both SDEs and sEEGs, higher frequency signals exhibited a sharper fall off relative to lower frequency signals. For broadband high γ (BHG) activity, the mean FWHM of SDEs (6.6 ± 2.5 mm) and sEEGs in gray matter (7.14 ± 1.7 mm) was not significantly different; however, FWHM for low frequencies recorded by sEEGs was 2.45 mm smaller than SDEs. White matter sEEGs showed much lower power for frequencies 17-200 Hz (q < 0.01) and a much broader decay (11.3 ± 3.2 mm) than gray matter electrodes (7.14 ± 1.7 mm). The use of a bipolar referencing scheme significantly lowered FWHM for sEEGs, relative to a white matter reference or a common average reference (CAR). These results outline the influence of array design, spectral bands, and referencing schema on local field potential recordings and source localization in icEEG recordings in humans. The metrics we derive have immediate relevance to the analysis and interpretation of both cognitive and epileptic data.

CHIME: CMOS-Hosted in vivo Microelectrodes for Massively Scalable Neuronal Recordings

Mammalian brains consist of 10s of millions to 100s of billions of neurons operating at millisecond time scales, of which current recording techniques only capture a tiny fraction. Recording techniques capable of sampling neural activity at high spatiotemporal resolution have been difficult to scale. The most intensively studied mammalian neuronal networks, such as the neocortex, show a layered architecture, where the optimal recording technology samples densely over large areas. However, the need for application-specific designs as well as the mismatch between the three-dimensional architecture of the brain and largely two-dimensional microfabrication techniques profoundly limits both neurophysiological research and neural prosthetics. Here, we discuss a novel strategy for scalable neuronal recording by combining bundles of glass-ensheathed microwires with large-scale amplifier arrays derived from high-density CMOS in vitro MEA systems or high-speed infrared cameras. High signal-to-noise ratio (<25 μV RMS noise floor, SNR up to 25) is achieved due to the high conductivity of core metals in glass-ensheathed microwires allowing for ultrathin metal cores (down to <1 μm) and negligible stray capacitance. Multi-step electrochemical modification of the tip enables ultra-low access impedance with minimal geometric area, which is largely independent of the core diameter. We show that the microwire size can be reduced to virtually eliminate damage to the blood-brain-barrier upon insertion and we demonstrate that microwire arrays can stably record single-unit activity. Combining microwire bundles and CMOS arrays allows for a highly scalable neuronal recording approach, linking the progress in electrical neuronal recordings to the rapid progress in silicon microfabrication. The modular design of the system allows for custom arrangement of recording sites. Our approach of employing bundles of minimally invasive, highly insulated and functionalized microwires to extend a two-dimensional CMOS architecture into the 3rd dimension can be translated to other CMOS arrays, such as electrical stimulation devices.

Nucleotide-time alignment for molecular recorders

Using a DNA polymerase to record intracellular calcium levels has been proposed as a novel neural recording technique, promising massive-scale, single-cell resolution monitoring of large portions of the brain. This technique relies on local storage of neural activity in strands of DNA, followed by offline analysis of that DNA. In simple implementations of this scheme, the time when each nucleotide was written cannot be determined directly by post-hoc DNA sequencing; the timing data must be estimated instead. Here, we use a Dynamic Time Warping-based algorithm to perform this estimation, exploiting correlations between neural activity and observed experimental variables to translate DNA-based signals to an estimate of neural activity over time. This algorithm improves the parallelizability of traditional Dynamic Time Warping, allowing several-fold increases in computation speed. The algorithm also provides a solution to several critical problems with the molecular recording paradigm: determining recording start times and coping with DNA polymerase pausing. The algorithm can generally locate DNA-based records to within <10% of a recording window, allowing for the estimation of unobserved incorporation times and latent neural tunings. We apply our technique to an in silico motor control neuroscience experiment, using the algorithm to estimate both timings of DNA-based data and the directional tuning of motor cortical cells during a center-out reaching task. We also use this algorithm to explore the impact of polymerase characteristics on system performance, determining the precision of a molecular recorder as a function of its kinetic and error-generating properties. We find useful ranges of properties for DNA polymerase-based recorders, providing guidance for future protein engineering attempts. This work demonstrates a useful general extension to dynamic alignment algorithms, as well as direct applications of that extension toward the development of molecular recorders, providing a necessary stepping stone for future biological work.

Spatial resolution dependence on spectral frequency in human speech cortex electrocorticography

OBJECTIVE

Electrocorticography (ECoG) has become an important tool in human neuroscience and has tremendous potential for emerging applications in neural interface technology. Electrode array design parameters are outstanding issues for both research and clinical applications, and these parameters depend critically on the nature of the neural signals to be recorded. Here, we investigate the functional spatial resolution of neural signals recorded at the human cortical surface. We empirically derive spatial spread functions to quantify the shared neural activity for each frequency band of the electrocorticogram.

APPROACH

Five subjects with high-density (4 mm center-to-center spacing) ECoG grid implants participated in speech perception and production tasks while neural activity was recorded from the speech cortex, including superior temporal gyrus, precentral gyrus, and postcentral gyrus. The cortical surface field potential was decomposed into traditional EEG frequency bands. Signal similarity between electrode pairs for each frequency band was quantified using a Pearson correlation coefficient.

MAIN RESULTS

The correlation of neural activity between electrode pairs was inversely related to the distance between the electrodes; this relationship was used to quantify spatial falloff functions for cortical subdomains. As expected, lower frequencies remained correlated over larger distances than higher frequencies. However, both the envelope and phase of gamma and high gamma frequencies (30-150 Hz) are largely uncorrelated (<90%) at 4 mm, the smallest spacing of the high-density arrays. Thus, ECoG arrays smaller than 4 mm have significant promise for increasing signal resolution at high frequencies, whereas less additional gain is achieved for lower frequencies.

SIGNIFICANCE

Our findings quantitatively demonstrate the dependence of ECoG spatial resolution on the neural frequency of interest. We demonstrate that this relationship is consistent across patients and across cortical areas during activity.

Towards large-scale, human-based, mesoscopic neurotechnologies

Direct human brain recordings have transformed the scope of neuroscience in the past decade. Progress has relied upon currently available neurophysiological approaches in the context of patients undergoing neurosurgical procedures for medical treatment. While this setting has provided precious opportunities for scientific research, it also has presented significant constraints on the development of new neurotechnologies. A major challenge now is how to achieve high-resolution spatiotemporal neural recordings at a large scale. By narrowing the gap between current approaches, new directions tailored to the mesoscopic (intermediate) scale of resolution may overcome the barriers towards safe and reliable human-based neurotechnology development, with major implications for advancing both basic research and clinical translation.