Deep learning and feature based medication classifications from EEG in a large clinical data set.

The amount of freely available human phenotypic data is increasing daily, and yet little is known about the types of inferences or identifying characteristics that could reasonably be drawn from that data using new statistical methods. One data type of particular interest is electroencephalographical (EEG) data, collected noninvasively from humans in various behavioral contexts. The Temple University EEG corpus associates thousands of hours of de-identified EEG records with contemporaneous physician reports that include metadata that might be expected to show a measurable correlation with characteristics of the recorded signal. Given that machine learning methods applied to neurological signals are being used in emerging diagnostic applications, we leveraged this data source to test the confidence with which algorithms could predict, using a patient's EEG record(s) as input, which medications were noted on the matching physician report. We comparatively assessed deep learning and feature-based approaches on their ability to distinguish between the assumed presence of Dilantin (phenytoin), Keppra (levetiracetam), or neither. Our methods could successfully distinguish between patients taking either anticonvulsant and those taking no medications; as well as between the two anticonvulsants. Further, we found different approaches to be most effective for different groups of classifications.

Assessing kinematic variability during performance of Jebsen-Taylor Hand Function Test.

STUDY DESIGN: Clinical measurement; 22 subjects with no upper limb disability completed the Jebsen-Taylor Hand Function Test (JHFT). INTRODUCTION: To realize the potential of 3D motion capture to augment evaluation of individuals with upper limb disability/impairment, it is important to understand the expected kinematic motion that characterizes performance during functional evaluation. PURPOSE OF THE STUDY: To assess kinematic variability and establish kinematic patterns for the JHFT. METHODS: Upper body joint kinematics were collected using a Vicon motion capture system. Average range of motion and maximum angle were calculated for all tasks. Intrasubject and intersubject variability were assessed by calculating Pearson's correlation coefficient, adjusted coefficient of multiple correlation (CMCadj), and standard deviation for 10 joint angles at the wrist, elbow, shoulder, and torso. RESULTS: The writing and picking up small objects tasks generally had high intrasubject variability, with most joint angles having median Pearson's correlation coefficients lower than 0.7. The CMCadj values were generally greater than 0.5 for elbow, shoulder, and torso joints during can-lifting tasks, indicating high consistency in those kinematic trajectories across subjects. Low consistency across subjects in all joint angles was observed for writing (CMCadj < 0.07; SDmax > 10°). DISCUSSION: Kinematic patterns for the JHFT tasks were analyzed. CONCLUSIONS: With kinematic patterns for the JHFT tasks analyzed, optimal patterns of activity performance can be defined, allowing for easier identification and adjustment of atypical motion. Results can be used to inform selection of tasks for kinematic evaluation and provide expected variability for comparison to patient populations, which is useful for regulatory review and clinical assessment.

Consistency of quantitative electroencephalography features in a large clinical data set.

OBJECTIVE: Despite their increasing use and public health importance, little is known about the consistency and variability of the quantitative features of baseline electroencephalography (EEG) measurements in healthy individuals and populations. This study aims to investigate population consistency of EEG features. APPROACH: We propose a non-parametric method of evaluating consistency of commonly used EEG features based on counts of non-significant statistical tests using a large data set. We first replicate stationarity results of absolute band powers using coefficients of variation. We then determine feature stationarity, intra-subject consistency, inter-subject consistency, and intra- versus inter-subject consistency across different epoch lengths for 30 features. MAIN RESULTS: We find in general that features with normalizing constants are more stationary. We also find entropy, median, skew, and kurtosis of EEG to behave as baseline EEG metrics. However, other spectral and signal shape features have stronger intra-subject consistency and thus are better for distinguishing individuals. SIGNIFICANCE: These results provide data-driven non-parametric methods of identifying EEG features and their spatial characteristics ideal for various EEG applications, and determining future EEG feature consistencies using an existing EEG data set.

Evaluation of Performance-Based Outcome Measures for the Upper Limb: A Comprehensive Narrative Review.

Objective performance-based outcome measures (OMs) have the potential to provide unbiased and reproducible assessments of limb function. However, very few of these performance-based OMs have been validated for upper limb (UL) prosthesis users. OMs validated in other clinical populations (eg, neurologic or musculoskeletal conditions) could be used to fill gaps in existing performance-based OMs for UL amputees. Additionally, a joint review might reveal consistent gaps across multiple clinical populations. Therefore, the objective of this review was to systematically characterize prominent measures used in both sets of clinical populations with regard to (1) location of task performance around the body, (2) possible grips employed, (3) bilateral versus unilateral task participation, and (4) details of scoring mechanisms. A systematic literature search was conducted in EMBASE, Medline, and Cumulative Index to Nursing and Allied Health electronic databases for variations of the following terms: stroke, musculoskeletal dysfunction, amputation, prosthesis, upper limb, outcome, assessments. Articles were included if they described performance-based OMs developed for disabilities of the UL. Results show most tasks were performed with 1 hand in the space directly in front of the participant. The tip, tripod, and cylindrical grips were most commonly used for the specific tasks. Few measures assessed sensation and movement quality. Overall, several limitations in OMs were identified. The solution to these limitations may be to modify and validate existing measures originally developed for other clinical populations as first steps to more aptly measure prosthesis use while more complete assessments for UL prosthesis users are being developed. LEVEL OF EVIDENCE: Level III.

An Integrated Movement Analysis Framework to Study Upper Limb Function: A Pilot Study.

The functional capabilities of individuals with upper limb disabilities are assessed throughout rehabilitation and treatment regimens using functional outcome measures. For the upper limb amputee population, there are none which quantitatively take into account the quality of movement while an individual is performing tasks. In this paper, we demonstrate the use of an integrated movement analysis framework, based on motion capture and ground reaction force data, to capture quantitative information about how subjects complete a commonly used functional outcome measure, the Box and Blocks Test (BBT). In order to test the usefulness of the integrated movement analysis framework in capturing the quality of movements during task performance, a motion restriction was induced in able-bodied participants that reproduces some of the limitations imposed by conventional prosthetics. Each subject performed the BBT under normal conditions and also under the motion restriction condition. The motion capture and ground force plates captured movement that significantly differed between the two conditions, with the largest differences seen in shoulder motion, in the range of motions of head tilt and elbow flexion, and in the area of the center of pressure trajectory. These preliminary results show the feasibility of incorporating standardized, quantitative movement analysis into the assessment of function for those with an upper limb disability.

Neuroprosthetics and the science of patient input.

Safe and effective neuroprosthetic systems are of great interest to both DARPA and CDRH, due to their innovative nature and their potential to aid severely disabled populations. By expanding what is possible in human-device interaction, these devices introduce new potential benefits and risks. Therefore patient input, which is increasingly important in weighing benefits and risks, is particularly relevant for this class of devices. FDA has been a significant contributor to an ongoing stakeholder conversation about the inclusion of the patient voice, working collaboratively to create a new framework for a patient-centered approach to medical device development. This framework is evolving through open dialogue with researcher and patient communities, investment in the science of patient input, and policymaking that is responsive to patient-centered data throughout the total product life cycle. In this commentary, we will discuss recent developments in patient-centered benefit-risk assessment and their relevance to the development of neural prosthetic systems.

Upper extremity prosthesis user perspectives on unmet needs and innovative technology.

The needs of individuals with upper limb amputation and congenital limb difference are not being fully met by current prostheses, as evidenced by prosthesis rejection, non-wear, and user reports of pain and challenging activities. Emerging technologies such as dexterous sensorized robotic limbs, osseointegrated prostheses, implantable EMG electrodes, and electrical stimulation for sensory feedback have the potential to address unmet needs, but pose additional risks. We plan to assess upper limb prosthesis user needs and perspectives on these new benefits and risks using an extensive quantitative survey. In preparation for this survey, we report here on qualitative interviews with seven individuals with upper limb amputation or congenital limb difference. Unstructured text was mined using topic modeling and the results compared with identified themes. A more complete understanding of how novel technologies could address real user concerns will inform implementation of new technologies and regulatory decision-making.

System to induce and measure embodiment of an artificial hand with programmable convergent visual and tactile stimuli.

The sense of prosthesis embodiment, or the feeling that the device has been incorporated into a user's body image, may be enhanced by emerging technology such as invasive electrical stimulation for sensory feedback. In turn, prosthesis embodiment may be linked to increased prosthesis use and improved functional outcomes. We describe the development of a tool to assay artificial hand embodiment in a quantitative way in people with intact limbs, and characterize its operation. The system delivers temporally coordinated visual and tactile stimuli at a programmable latency while recording limb temperature. When programmed to deliver visual and tactile stimuli synchronously, recorded latency between the two was 33 ± 24 ms in the final pilot subject. This system enables standardized assays of the conditions necessary for prosthesis embodiment.

Calcium-based dendritic excitability and its regulation in the deep cerebellar nuclei.

The deep cerebellar nuclei (DCN) convey the final output of the cerebellum and are a major site of activity-dependent plasticity. Here, using patch-clamp recording and two-photon calcium imaging in rat brain slices, we demonstrate that DCN dendrites exhibit three hallmarks of active amplification of electrical signals. First, they produce calcium transients with rise times of tens of milliseconds, comparable in amplitude and duration to calcium spikes in other neurons. Second, calcium signal amplitudes are undiminished along the length of dendrites to the farthest distances from the soma. Third, they can generate calcium signals even in the presence of tetrodotoxin, a sodium channel blocker that abolishes somatic action potential initiation. DCN calcium transients do require the action of T-type calcium channels, a common voltage-gated conductance in excitable dendrites. Dendritic calcium influx was evoked by release from hyperpolarization, peaked within tens of milliseconds, and was observed in both transient- and weak-rebound-firing neurons. In a survey across the DCN, transient-burst rebound firing, which was accompanied by the most rapid calcium flux, was more common in lateral nucleus than in interpositus nucleus and was not seen in medial nucleus. Rebound firing and calcium transients were not present in animals shipped 1-3 days before recording, a condition associated with elevated maternal and pup corticosterone and reduced pup body weight. Rebounds could be restored by the protein kinase C activator phorbol 12-myristate-13-acetate. Thus local calcium-based dendritic excitability supports a stage of presomatic amplification that is under regulation by stress and neuromodulatory influence.

Construction, alignment, and implementation of an acousto-optical deflector-based system for patterned uncaging with ultraviolet light.

The method of patterned photoactivation is a natural fit for the study of neuronal dendritic integration. Photoactivatable molecules that influence a wide range of extracellular and intracellular neurophysiological functions are available. The choice of photosensitive molecules depends on the research question and will influence the design of the experimental apparatus. An acousto-optical deflector (AOD)-based system can be used for rapid ultraviolet (UV) photolysis in arbitrary spatial and temporal patterns. Photolysis-activated "caged" diffusible molecules or newer light-sensitive membrane proteins can be used in this system. This protocol describes the addition of a UV beam for uncaging to a homebuilt two-photon microscope. The goal is to get UV light from the light source (laser) to the approximate center of the objective's back aperture, passing through a pair of perpendicularly oriented AODs along the way. The protocol also describes the fine alignment of the UV beam and the implementation of AOD-based beam steering. Performing the final alignment with the beam passing through the AODs will ensure that the system is optimized for the idiosyncrasies of the crystals.

Acousto-optical deflector-based patterned ultraviolet uncaging of neurotransmitter for the study of neuronal integration.

The method of patterned photoactivation is a natural fit for the study of neuronal dendritic integration. Photoactivatable molecules that influence a wide range of extracellular and intracellular neurophysiological functions are available. The choice of photosensitive molecules depends on the research question and will influence the design of the experimental apparatus. This article describes an acousto-optical deflector (AOD)-based system for rapid ultraviolet (UV) photolysis in arbitrary spatial and temporal patterns. Some basics of caged neurotransmitters and the theory of operation of AODs are covered, as are descriptions for implementing the system.

Preparing, handling, and applying caged compound solutions for acousto-optical deflector-based patterned uncaging with ultraviolet light.

The method of patterned photoactivation is a natural fit for the study of neuronal dendritic integration. Photoactivatable molecules that influence a wide range of extracellular and intracellular neurophysiological functions are available. The choice of photosensitive molecules depends on the research question and will influence the design of the experimental apparatus. An acousto-optical deflector (AOD)-based system can be used for rapid ultraviolet (UV) photolysis in arbitrary spatial and temporal patterns. This protocol describes how to prepare caged neurotransmitter compound solutions for use in this system and discusses options for introducing caged compounds into an experimental preparation.

Spatiotemporal properties of sensory responses in vivo are strongly dependent on network context.

Sensory responses in neocortex are strongly modulated by changes in brain state, such as those observed between sleep stages or attentional levels. However, the specific effects of network state changes on the spatiotemporal properties of sensory responses are poorly understood. The slow oscillation, which is observed in neocortex under ketamine-xylazine anesthesia and is characterized by alternating depolarizing (up-states) and hyperpolarizing (down-states) phases, provides an opportunity to study the state-dependence of primary sensory responses in large networks. Here we used voltage sensitive dye (VSD) imaging to record the spatiotemporal properties of sensory responses and local field potential (LFP) and multiunit activity (MUA) recordings to monitor the ongoing brain state in which the sensory responses occurred. Despite a rich variability of slow oscillation patterns, sensory responses showed a consistent relationship with the ongoing oscillation and triggered a new up-state only after the termination of the refractory period that followed the preceding oscillatory cycle. We show that spatiotemporal properties of whisker-evoked responses are highly dependent on their timing with regard to the ongoing oscillation. In both the up- and down-states, responses spread across large portions of the barrel field, although the up-state responses were reduced in total area due to their sparseness. The depolarizing response in the up-state showed a tendency to propagate along the rows, with an amplitude and slope favoring the higher-numbered arcs. In the up-state, but not in the down-state, the depolarizing response was followed by a hyperpolarizing wave with a consistent spatial structure. We measured the suppression of whisker-evoked responses by a preceding response at 100 ms, and found that suppression showed the same spatial asymmetry as the depolarization. Because the resting level of cells in the up-state is likely to be closer to that in the awake animal, we suggest that the polarities in signal propagation which we observed in the up-state could be used as computational mechanisms in the behaving animal. These results demonstrate the critical importance of ongoing network activity on the dynamics of sensory responses and their integration.

Integration of evoked responses in supragranular cortex studied with optical recordings in vivo.

Complex representations in sensory cortices rely on the integration of inputs that overlap temporally and spatially, particularly in supragranular layers, yet the spatiotemporal dynamics of this synaptic integration are largely unknown. The rodent somatosensory system offers an excellent opportunity to study these dynamics because of the overlapping functional representations of single-whisker inputs. We recorded responses in mouse primary somatosensory (barrel) cortex to single and paired whisker deflections using high-speed voltage-sensitive dye imaging. Responses to paired deflections at intervals of 0 and 10 ms summed sublinearly, producing a single transient smaller in amplitude than the sum of the component responses. At longer intervals of 50 and 100 ms, the response to the second deflection was reduced in amplitude and limited spatially relative to control. Between 100 and 200 ms, the response to the second deflection recovered and often showed areas of facilitation. With increasing interstimulus interval from 50 to 200 ms, recovery of the second response occurred from the second stimulated whisker's barrel column outward. In contrast to results with paired-whisker stimulation, when a whisker deflection was preceded by a weak electrical stimulus applied to the neighboring cortex, the summation of evoked responses was predominantly linear at all intervals tested. Thus under our conditions, the linearity of response summation in cortex was not predicted by the amplitudes of the component responses on a column-by-column basis, but rather by the timing and nature of the inputs.

In vivo fluorescence microscopy of neuronal activity in three dimensions by use of voltage-sensitive dyes.

We report in vivo imaging of neuronal electrical activity from superficial layers of the mouse barrel cortex. The measurements have approximately 16-microm spatial and 3-ms temporal resolution and reach depths of 150 microm below the cortical surface. The depth-dependent differential-fluorescence optical sections of activity are consistent with known cortical architecture and represent an important step toward in vivo measurement of functioning complex neural networks. Our observations employ a custom gradient-index lens probe and voltage-sensitive dye fluorescence; the use of epi-illumination rather than dark-field illumination provides the dramatic signal-to-noise improvement necessary for fast three-dimensional imaging.