Online detection and sorting of extracellularly recorded action potentials in human medial temporal lobe recordings, in vivo

Understanding the function of complex cortical circuits requires the simultaneous recording of action potentials from many neurons in awake and behaving animals. Practically, this can be achieved by extracellularly recording from multiple brain sites using single wire electrodes. However, in densely packed neural structures such as the human hippocampus, a single electrode can record the activity of multiple neurons. Thus, analytic techniques that differentiate action potentials of different neurons are required. Offline spike sorting approaches are currently used to detect and sort action potentials after finishing the experiment. Because the opportunities to record from the human brain are relatively rare, it is desirable to analyze large numbers of simultaneous recordings quickly using online sorting and detection algorithms. In this way, the experiment can be optimized for the particular response properties of the recorded neurons. Here we present and evaluate a method that is capable of detecting and sorting extracellular single-wire recordings in realtime. We demonstrate the utility of the method by applying it to an extensive data set we acquired from chronically implanted depth electrodes in the hippocampus of human epilepsy patients. This dataset is particularly challenging because it was recorded in a noisy clinical environment. This method will allow the development of "closed-loop" experiments, which immediately adapt the experimental stimuli and/or tasks to the neural response observed.     

An oscilloscope spot intensifier, to improve photographic recordings of action potentials

A circuit diagram is shown for a semiconductor device to intensify the brightness of an oscilloscope during the rapidly rising and falling phases of signals such as action potentials. Brightening pulses proportional in amplitude to the rate of change in the Y-axis are available for connection to an oscilloscope with an external intensity ('Z') modulation input. The circuit requires one transistor, one dual operational amplifier and two single fast operational amplifiers.     

DATA-MEAns: an open source tool for the classification and management of neural ensemble recordings

The number of laboratories using techniques that allow to acquire simultaneous recordings of as many units as possible is considerably increasing. However, the development of tools used to analyse this multi-neuronal activity is generally lagging behind the development of the tools used to acquire these data. Moreover, the data exchange between research groups using different multielectrode acquisition systems is hindered by commercial constraints such as exclusive file structures, high priced licenses and hard policies on intellectual rights. This paper presents a free open-source software for the classification and management of neural ensemble data. The main goal is to provide a graphical user interface that links the experimental data to a basic set of routines for analysis, visualization and classification in a consistent framework. To facilitate the adaptation and extension as well as the addition of new routines, tools and algorithms for data analysis, the source code and documentation are freely available.     

Real-time human action classification using a dynamic neural model

The multiple timescale recurrent neural network (MTRNN) model is a useful tool for recording and regenerating a continuous signal for dynamic tasks. However, our research shows that the MTRNN model is difficult to use for the classification of multiple types of motion when observing a human action. Therefore, in this paper, we propose a new supervised MTRNN model for handling the issue of action classification. Instead of setting the initial states, we define a group of slow context nodes as “classification nodes.” The supervised MTRNN model provides both prediction and classification outputs simultaneously during testing. Our experiment results show that the supervised MTRNN model inherits the basic function of an MTRNN and can be used to generate action signals. In addition, the results show that the robustness of the supervised MTRNN model is better than that of the MTRNN model when generating both action sequences and action classification tasks.     

The optimal recording electrode configuration for compound sensory action potentials.

There is no uniformity in the published literature from different laboratories on the optimal electrode configuration for recording nerve action potentials, and a number of standard texts omit any reference to the effects that interelectrode distance and electrode orientation can have on the shape, amplitude and latency of nerve action potentials. The sensory action potential from the digital nerves of the index finger was recorded at wrist and elbow using bipolar electrodes with the "active" electrode over the median nerve and the "reference" placed 4 cm laterally or proximally along the nerve using interelectrode distances of 4, 3 and 2 cm. These potentials were compared with that recorded using a remote reference on the ipsilateral shoulder, the assumption being that this configuration eliminated the contribution of the reference electrode to the compound nerve action potential. With different electrode configurations, there were significant differences in the shape of the potential, the latencies to onset and peak and the rising- and falling-phase amplitudes. The shorter the distance between the electrodes the greater the distortions. Overall, the distortions were least with the 4 cm interelectrode separation, particularly for short conduction distances.     

Automatic spike detection based on adaptive template matching for extracellular neural recordings

Recordings of extracellular neural activity are used in many clinical applications and scientific studies. In most cases, these signals are analyzed as a point process, and a spike detection algorithm is required to estimate the times at which action potentials occurred. Recordings from high-density microelectrode arrays (MEAs) and low-impedance microelectrodes often have a low signal-to-noise ratio (SNR<10) and contain action potentials from more than one neuron. We describe a new detection algorithm based on template matching that only requires the user to specify the minimum and maximum firing rates of the neurons. The algorithm iteratively estimates the morphology of the most prominent action potentials. It is able to achieve a sensitivity of >90% with a false positive rate of <5Hz in recordings with an estimated SNR=3, and it performs better than an optimal threshold detector in recordings with an estimated SNR>2.5.     

Automated interictal EEG spike detection using artificial neural networks.

Feed-forward, error-back-propagation artificial neural networks were applied to recognition of epileptiform patterns in the EEG. The inherent network properties of generalization and variability tolerance were effective in identifying wave forms that differed from the training patterns but still maintained 'epileptiform' spatio-temporal characteristics. The certainty of recognition was measured as a continuous function with a range of 0-1. Two levels of certainty (0.825 and 0.900) were used to indicate recognition of spikes and sharp waves (SSW). An average 94.2% (+/- 7.3) of the SSW were recognized; 20.9% (+/- 22.9) of all recognized SSW were false-positive recognitions. The time required for pattern recognition was well within the time required for digitizing the analogue data. This study provides evidence that neural network technology is, in principle, an effective pattern recognition strategy for identification of epileptiform transients in the EEG. The analysis is sufficiently rapid to be of potential value as a strategy for data reduction of long recordings stored on bulk media.     

Sucrose-gap recordings of nerve-evoked potentials in mammalian parasympathetic ganglia

The sucrose-gap recording technique was used to study mammalian parasympathetic ganglionic transmission. Both a fast nicotinic depolarization potential and a slow muscarinic hyperpolarizing potential were recorded in vesical pelvic ganglia (VPG). A long afterhyperpolarization caused by an electrical response of through-fibers was also recorded. However, a slow excitatory postsynaptic potential (S-EPSP) was not readily observed and may be masked by the long afterhyperpolarization. In addition, the slow inhibitory postsynaptic potential (S-IPSP) of the VPG was due to a direct effect of acetylcholine (ACh). Thus, sucrose-gap recordings of VPG potentials are similar to those obtained in sympathetic ganglia, but the mechanism for transmission of the S-IPSP may be different in the respective ganglia.     

Scalp and depth recordings of induced deep cerebral potentials

A balanced square wave was introduced between two adjacent depth electrodes implanted in the course of studying patients with intractable epilepsy and who were being considered for surgery. The stimulus current was designed so that charge density loading was well within limits of safety to avoid tissue damage. No neuronal activation was seen, and the stimulus intensity was significantly less than that used in subsequent stimulation session for the purpose of eliciting a clinical response and after-discharges. Averaging techniques were used to record the stimulus at distant electrodes both within the cerebrum and on the scalp. The recorded voltage decrement from the source was nearly identical with the theoretical voltage decrement predicted using principles of electric field theory in which the brain was assumed to be a homogeneous conductive medium. When the voltage recorded on the scalp was compared with the voltages recorded from depth electrodes, it was found that the effect of the highly resistive skull on voltage decrement was relatively less the more centric the source. This result also confirmed predictions based on electric field theory. Most significantly, voltages well within the physiologic range introduced in deep mesial temporal lobe structures were recorded from the scalp.     

Methods for isolating extracellular action potentials and removing stimulus artifacts from microelectrode recordings of neurons requiring minimal operator intervention

Recent successes in treating neurological disorders with electrical stimulation of the brain have spurred interest in studying the neuronal mechanisms by which such therapies work. However, microelectrode recordings can be confounded by stimulation artifact. Also, large microelectrode arrays now allow recording amounts of data that would otherwise overwhelm current analytic methods that depend heavily on human intervention and interpretation. A set of algorithms is described for automatically removing stimulus artifacts that minimize signal loss with minimum human involvement. Other algorithms automatically differentiate between the extracellular action potentials of individual neurons.     

Important abnormalities in recordings of somatosensory evoked potentials in coma

We have chosen six illustrations showing how much vital information can be obtained from median nerve SEPs during the first 24 hours in coma. With avulsion of brachial plexus roots there was loss of SEPs at the cervical cord and the scalp from the affected side. In a severe injury of the cervical cord there was preservation of brachial plexus potentials, while SEPs at the cervical cord were absent. After critical deterioration in a case of repeated subarachnoidal hemorrhage, scalp SEPs with very short latency occurred, which is a finding suggestive of destruction of cortical SEP generators heralding a fatal outcome. In a case of brain injury combined with central hyperthermia, there was initially a loss of scalp SEPs probably due to the combined effect of these factors. In a case of brain injury there were bifid peaks at the scalp level. It is important to assess central sensory conduction time only to the first scalp SEP, otherwise an erroneously abnormal state may be inferred. In a patient with clinical and EEG evidence of brain death there was a loss of far-field thalamic potentials at the neck. It is important to be aware of such presentations to be able to provide corroborative assurance for the assessment of prognosis.     

Intracranial recordings of movement-related potentials to voluntary saccades

Movement-related potentials evoked by voluntary and self-initiated horizontal saccades were recorded from subdural electrodes placed over the lateral (premotor and motor cortex) and the mesial (supplementary motor area) surfaces of the frontal lobe, in four patients with intractable focal seizures. An extremely localized bereitschaftspotential showing approximately the same latencies and amplitudes was simultaneously recorded from the frontal eye field and supplementary motor area (SMA). Our data suggest that both regions are equally active prior to saccades and do not support the view that the SMA acts as a supramotor cortex, being activated during the planning of the movement and the primary motor cortex only later on, close to the execution of the movement. In addition, we never observed the spike potential in our intracranial recordings, thus supporting the hypothesis of its extracerebral origin.     

Line length as a robust method to detect high-activity events: Automated burst detection in premature EEG recordings

Objective

EEG is a valuable tool for evaluation of brain maturation in preterm babies. Preterm EEG constitutes of high voltage burst activities and more suppressed episodes, called interburst intervals (IBIs). Evolution of background characteristics provides information on brain maturation and helps in prediction of neurological outcome. The aim is to develop a method for automated burst detection.

Methods

Thirteen polysomnography recordings were used, collected at preterm postmenstrual age of 31.4 (26.1–34.4) weeks. We developed a burst detection algorithm based on the feature line length and compared it with manual scorings of clinical experts and other published methods.

Results

The line length–based algorithm is robust (84.27% accuracy, 84.00% sensitivity, 85.70% specificity). It is not critically dependent on the number of measurement channels, because two channels still provide 82% accuracy. Furthermore, it approximates well clinically relevant features, such as median IBI duration 5.45 (4.00–7.11) s, maximum IBI duration 14.02 (8.73–18.80) s and burst percentage 48.89 (35.45–60.12)%, with a median deviation of respectively 0.65 s, 1.96 s and 6.55%.

Conclusion

Automated assessment of long-term preterm EEG is possible and its use will optimize EEG interpretation in the NICU.

Significance

This study takes a first step towards fully automatic analysis of the preterm brain.     

Minocycline increases quality and longevity of chronic neural recordings

Brain/machine interfaces could potentially be used in the treatment of a host of neurological disorders ranging from paralysis to sensory deficits. Insertion of chronic micro-electrode arrays into neural tissue initiates a host of immunological responses, which typically leads to the formation of a cellular sheath around the implant, resulting in the loss of useful signals. Minocycline has been shown to have neuroprotective and neurorestorative effects in certain neural injury and neurodegenerative disease models. This study examined the effects of minocycline administration on the quality and longevity of chronic multi-channel microwire neural implants 1 week and 1 month post-implantation in auditory cortex. The mean signal-to-noise ratio for the minocycline group stabilized at the end of week 1 and remained above 4.6 throughout the following 3 weeks. The control group signal-to-noise ratio dropped throughout the duration of the study and at the end of 4 weeks was 2.6. Furthermore, 68% of electrodes from the minocycline group showed significant stimulus-driven activity at week 4 compared to 12.5% of electrodes in the control group. There was a significant reduction in the number of activated astrocytes around the implant in minocycline subjects, as well as a reduction in total area occupied by activated astrocytes at 1 and 4 weeks.     

Automated classification of evoked quantal events

We provide both theoretical and computational improvements to the analysis of synaptic transmission data. Theoretically, we demonstrate the correlation structure of observations within evoked postsynaptic potentials (EPSP) are consistent with multiple random draws from a common autoregressive moving-average (ARMA) process of order (2, 2). We use this observation and standard time series results to construct a statistical hypothesis testing procedure for determining whether a given trace is an EPSP. Computationally, we implement this method in R, a freeware statistical language, which reduces the amount of time required for the investigator to classify traces into EPSPs or non-EPSPs and eliminates investigator subjectivity from this classification. In addition, we provide a computational method for calculating common functionals of EPSPs (peak amplitude, decay rate, etc.). The methodology is freely available over the internet. The automated procedure to index the quantal characteristics greatly facilitates determining if any one or multiple parameters are changing due to experimental conditions. In our experience, the software reduces the time required to perform these analyses from hours to minutes.     

Accuracy and inter-observer variation in the classification of dysarthria from speech recordings

Background   Dysarthria may be classified as flaccid, spastic, ataxic, hypokinetic, choreatic, dystonic, or mixed. We hypothesized that in routine neurological practice the reliability and accuracy of perceptual analysis alone in the classification of dysarthria is low and that this classification is mainly based on the clinical context rather than on the perception of speech. We therefore studied the accuracy and the inter- observer agreement in the classification of dysarthrias on the basis of perceptual analysis alone. Methods   Seventy two neurologists and neurological trainees classified recorded speech samples of 100 patients as flaccid, spastic, ataxic, extrapyramidal, or mixed dysarthria, or as not dysarthric. All observers were blinded to the patients’ final diagnosis, which was based on all clinical features and investigations. In the analysis the observers were arranged in eight groups of nine observers, or four paired groups with similar levels of clinical experience. Together, the observers in a given group rated all 100 recordings. Results   The accuracy of the classification was poor (35 % were classified correctly) and the inter-observer agreement between paired groups low (κ 0.16 to 0.32). The level of experience in neurology did not have a significant influence. Conclusion   Neurological trainees as well as experienced neurologists have great difficulty in identifying specific types of dysarthria on the basis of perceptual analysis alone. In clinical practice this probably means that most neurologists will classify dysarthria in the context of other features from neurological examination or ancillary investigations.     

Single trial classification of magnetoencephalographic recordings using Granger causality

The use of Granger causality (GC) for studying dependencies in neuroimaging data has recently been gaining popularity. Several frameworks exist for applying GC to neurophysiological questions but many rely heavily on specific statistical assumptions regarding autoregressive (AR) models for hypothesis testing. Since it is often difficult to satisfy these assumptions in practical settings, this study proposes an alternative statistical methodology based on the classification of individual trials of data. Instead of testing for significance using statistics based on estimated AR models or prediction errors, hypotheses were tested by determining whether or not individual magnetoencephalography (MEG) recording segments belonging to either of two experimental conditions can be successfully classified using features derived from AR and GC concepts. Using this novel approach, we show that bivariate temporal GC can be used to distinguish button presses based on whether they were experimentally forced or free. Additionally, the methodology was used to determine useful parameter settings for various steps of the analysis and this revealed surprising insight into several aspects of AR and GC analysis which, previously, could not be obtained in a comparable manner. A final mean accuracy of 79.2% was achieved for classifying forced and free button presses for 6 subjects suggesting that classification using GC features is a viable option for studying MEG signals and useful for evaluating the effectiveness of parameter variations in GC analysis.