Dissecting the microvascular contributions to diffuse correlation spectroscopy measurements of cerebral hemodynamics using optical coherence tomography angiography.

Significance: Diffuse correlation spectroscopy (DCS) is an emerging noninvasive, diffuse optical modality that purportedly enables direct measurements of microvasculature blood flow. Functional optical coherence tomography angiography (OCT-A) can resolve blood flow in vessels as fine as capillaries and thus has the capability to validate key attributes of the DCS signal. Aim: To characterize activity in cortical vasculature within the spatial volume that is probed by DCS and to identify populations of blood vessels that are most representative of the DCS signals. Approach: We performed simultaneous measurements of somatosensory-evoked cerebral blood flow in mice in vivo using both DCS and OCT-A. Results: We resolved sensory-evoked blood flow in the somatosensory cortex with both modalities. Vessels with diameters smaller than 10 µ m featured higher peak flow rates during the initial poststimulus positive increase in flow, whereas larger vessels exhibited considerably larger magnitude of the subsequent undershoot. The simultaneously recorded DCS waveforms correlated most highly with flow in the smallest vessels, yet featured a more prominent undershoot. Conclusions: Our direct, multiscale, multimodal cross-validation measurements of functional blood flow support the assertion that the DCS signal preferentially represents flow in microvasculature. The significantly greater undershoot in DCS, however, suggests a more spatially complex relationship to flow in cortical vasculature during functional activation.

Evaluating Performance of EEG Data-Driven Machine Learning for Traumatic Brain Injury Classification.

OBJECTIVES: Big data analytics can potentially benefit the assessment and management of complex neurological conditions by extracting information that is difficult to identify manually. In this study, we evaluated the performance of commonly used supervised machine learning algorithms in the classification of patients with traumatic brain injury (TBI) history from those with stroke history and/or normal EEG. METHODS: Support vector machine (SVM) and K-nearest neighbors (KNN) models were generated with a diverse feature set from Temple EEG Corpus for both two-class classification of patients with TBI history from normal subjects and three-class classification of TBI, stroke and normal subjects. RESULTS: For two-class classification, an accuracy of 0.94 was achieved in 10-fold cross validation (CV), and 0.76 in independent validation (IV). For three-class classification, 0.85 and 0.71 accuracy were reached in CV and IV respectively. Overall, linear discriminant analysis (LDA) feature selection and SVM models consistently performed well in both CV and IV and for both two-class and three-class classification. Compared to normal control, both TBI and stroke patients showed an overall reduction in coherence and relative PSD in delta frequency, and an increase in higher frequency (alpha, mu, beta and gamma) power. But stroke patients showed a greater degree of change and had additional global decrease in theta power. CONCLUSIONS: Our study suggests that EEG data-driven machine learning can be a useful tool for TBI classification. SIGNIFICANCE: Our study provides preliminary evidence that EEG ML algorithm can potentially provide specificity to separate different neurological conditions.

Longitudinal multimodal assessment of neurodegeneration and vascular remodeling correlated with signal degradation in chronic cortical silicon microelectrodes.

Significance: Cortically implanted microelectrode arrays provide a direct interface with neuronal populations and are used to restore movement capabilities and provide sensory feedback to patients with paralysis or amputation. Penetrating electrodes experience high rates of signal degradation within the first year that limit effectiveness and lead to eventual device failure. Aim: To assess vascular and neuronal changes over time in mice with implanted electrodes and examine the contribution of the brain tissue response to electrode performance. Approach: We used a multimodal approach combining in vivo electrophysiology and subcellular-level optical imaging. Results: At acute timescales, we observed structural damage from the mechanical trauma of electrode insertion, evidenced by severed dendrites in the electrode path and local hypofluorescence. Superficial vessel growth and remodeling occurred within the first few weeks in both electrode-implanted and window-only animals, but the deeper capillary growth evident in window-only animals was suppressed in electrode-implanted animals. After longer implantation periods, there was evidence of degeneration of transected dendrites superficial to the electrode path and localized neuronal cell body loss, along with deep vascular velocity changes near the electrode. Total spike rate (SR) across all animals reached a peak between 3 and 9 months postimplantation, then decreased. The local field potential signal remained relatively constant for up to 6 months, particularly in the high-gamma band, indicating long-term electrode viability and neuronal functioning at further distances from the electrode, but it showed a reduction in some animals at later time points. Most importantly, we found that progressive high-gamma and SR reductions both correlate positively with localized cell loss and decreasing capillary density within 100 µ m of the electrode. Conclusions: This multifaceted approach provided a more comprehensive picture of the ongoing biological response at the brain-electrode interface than can be achieved with postmortem histology alone and established a real-time relationship between electrophysiology and tissue damage.

Transient Subgranular Hyperconnectivity to L2/3 and Enhanced Pairwise Correlations During the Critical Period in the Mouse Auditory Cortex.

During the critical period, neuronal connections are shaped by sensory experience. While the basis for this temporarily heightened plasticity remains unclear, shared connections introducing activity correlations likely play a key role. Thus, we investigated the changing intracortical connectivity in primary auditory cortex (A1) over development. In adult, layer 2/3 (L2/3) neurons receive ascending inputs from layer 4 (L4) and also receive few inputs from subgranular layer 5/6 (L5/6). We measured the spatial pattern of intracortical excitatory and inhibitory connections to L2/3 neurons in slices of mouse A1 across development using laser-scanning photostimulation. Before P11, L2/3 cells receive most excitatory input from within L2/3. Excitatory inputs from L2/3 and L4 increase after P5 and peak during P9-16. L5/6 inputs increase after P5 and provide most input during P12-16, the peak of the critical period. Inhibitory inputs followed a similar pattern. Functional circuit diversity in L2/3 emerges after P16. In vivo two-photon imaging shows low pairwise signal correlations in neighboring neurons before P11, which peak at P15-16 and decline after. Our results suggest that the critical period is characterized by high pairwise activity correlations and that transient hyperconnectivity of specific circuits, in particular those originating in L5/6, might play a key role.

Temporary Visual Deprivation Causes Decorrelation of Spatiotemporal Population Responses in Adult Mouse Auditory Cortex.

Although within-modality sensory plasticity is limited to early developmental periods, cross-modal plasticity can occur even in adults. In vivo electrophysiological studies have shown that transient visual deprivation (dark exposure, DE) in adult mice improves the frequency selectivity and discrimination of neurons in thalamorecipient layer 4 (L4) of primary auditory cortex (A1). Since sound information is processed hierarchically in A1 by populations of neurons, we investigated whether DE alters network activity in A1 L4 and layer 2/3 (L2/3). We examined neuronal populations in both L4 and L2/3 using in vivo two-photon calcium (Ca2+) imaging of transgenic mice expressing GCaMP6s. We find that one week of DE in adult mice increased the sound evoked responses and frequency selectivity of both L4 and L2/3 neurons. Moreover, after DE the frequency representation changed with L4 and L2/3 showing a reduced representation of cells with best frequencies (BFs) between 8 and 16 kHz and an increased representation of cells with BFs above 32 kHz. Cells in L4 and L2/3 showed decreased pairwise signal correlations (SCs) consistent with sharper tuning curves. The decreases in SCs were larger in L4 than in L2/3. The decreased pairwise correlations indicate a sparsification of A1 responses to tonal stimuli. Thus, cross-modal experience in adults can both alter the sound-evoked responses of A1 neurons and change activity correlations within A1 potentially enhancing the encoding of auditory stimuli.

Longitudinal Functional Assessment of Brain Injury Induced by High-Intensity Ultrasound Pulse Sequences.

Exposure of the brain to high-intensity stress waves creates the potential for long-term functional deficits not related to thermal or cavitational damage. Possible sources of such exposure include overpressure from blast explosions or high-intensity focused ultrasound (HIFU). While current ultrasound clinical protocols do not normally produce long-term neurological deficits, the rapid expansion of potential therapeutic applications and ultrasound pulse-train protocols highlights the importance of establishing a safety envelope beyond which therapeutic ultrasound can cause neurological deficits not detectable by standard histological assessment for thermal and cavitational damage. In this study, we assessed the neuroinflammatory response, behavioral effects, and brain micro-electrocorticographic (µECoG) signals in mice following exposure to a train of transcranial pulses above normal clinical parameters. We found that the HIFU exposure induced a mild regional neuroinflammation not localized to the primary focal site, and impaired locomotor and exploratory behavior for up to 1 month post-exposure. In addition, low frequency (δ) and high frequency (ß, γ) oscillations recorded by ECoG were altered at acute and chronic time points following HIFU application. ECoG signal changes on the hemisphere ipsilateral to HIFU exposure are of greater magnitude than the contralateral hemisphere, and persist for up to three months. These results are useful for describing the upper limit of transcranial ultrasound protocols, and the neurological sequelae of injury induced by high-intensity stress waves.

Fast and Stable Signal Deconvolution via Compressible State-Space Models.

OBJECTIVE: Common biological measurements are in the form of noisy convolutions of signals of interest with possibly unknown and transient blurring kernels. Examples include EEG and calcium imaging data. Thus, signal deconvolution of these measurements is crucial in understanding the underlying biological processes. The objective of this paper is to develop fast and stable solutions for signal deconvolution from noisy, blurred, and undersampled data, where the signals are in the form of discrete events distributed in time and space. METHODS: We introduce compressible state-space models as a framework to model and estimate such discrete events. These state-space models admit abrupt changes in the states and have a convergent transition matrix, and are coupled with compressive linear measurements. We consider a dynamic compressive sensing optimization problem and develop a fast solution, using two nested expectation maximization algorithms, to jointly estimate the states as well as their transition matrices. Under suitable sparsity assumptions on the dynamics, we prove optimal stability guarantees for the recovery of the states and present a method for the identification of the underlying discrete events with precise confidence bounds. RESULTS: We present simulation studies as well as application to calcium deconvolution and sleep spindle detection, which verify our theoretical results and show significant improvement over existing techniques. CONCLUSION: Our results show that by explicitly modeling the dynamics of the underlying signals, it is possible to construct signal deconvolution solutions that are scalable, statistically robust, and achieve high temporal resolution. SIGNIFICANCE: Our proposed methodology provides a framework for modeling and deconvolution of noisy, blurred, and undersampled measurements in a fast and stable fashion, with potential application to a wide range of biological data.