Get a New Perspective on EEG: Convolutional Neural Network Encoders for Parametric t-SNE.

t-distributed stochastic neighbor embedding (t-SNE) is a method for reducing high-dimensional data to a low-dimensional representation, and is mostly used for visualizing data. In parametric t-SNE, a neural network learns to reproduce this mapping. When used for EEG analysis, the data are usually first transformed into a set of features, but it is not known which features are optimal. The principle of t-SNE was used to train convolutional neural network (CNN) encoders to learn to produce both a high- and a low-dimensional representation, eliminating the need for feature engineering. To evaluate the method, the Temple University EEG Corpus was used to create three datasets with distinct EEG characters: (1) wakefulness and sleep; (2) interictal epileptiform discharges; and (3) seizure activity. The CNN encoders produced low-dimensional representations of the datasets with a structure that conformed well to the EEG characters and generalized to new data. Compared to parametric t-SNE for either a short-time Fourier transform or wavelet representation of the datasets, the developed CNN encoders performed equally well in separating categories, as assessed by support vector machines. The CNN encoders generally produced a higher degree of clustering, both visually and in the number of clusters detected by k-means clustering. The developed principle is promising and could be further developed to create general tools for exploring relations in EEG data.

Investigations into an overlooked early component of painful nociceptive withdrawal reflex responses in humans.

Introduction: The role of pain as a warning system necessitates a rapid transmission of information from the periphery for the execution of appropriate motor responses. The nociceptive withdrawal reflex (NWR) is a physiological response to protect the limb from a painful stimulus and is often considered an objective measure of spinal nociceptive excitability. The NWR is commonly defined by its latency in the presumed Aδ-fiber range consistent with the canonical view that "fast pain" is signaled by Aδ nociceptors. We recently demonstrated that human skin is equipped with ultrafast (Aß range) nociceptors. Here, we investigated the short-latency component of the reflex and explored the relationship between reflex latency and pain perception. Methods: We revisited our earlier work on NWR measurements in which, following convention, only reflex responses in the presumed Aδ range were considered. In our current analysis, we expanded the time window to search for shorter latency responses and compared those with pain ratings. Results: In both cohorts, we found an abundance of recordings with short-latency reflex responses. In nearly 90% of successful recordings, only single reflex responses (not dual) were seen which allowed us to compare pain ratings based on reflex latencies. We found that shorter latency reflexes were just as painful as those in the conventional latency range. Conclusion: We found a preponderance of short-latency painful reflex responses. Based on this finding, we suggest that short-latency responses must be considered in future studies. Whether these are signaled by the ultrafast nociceptors remains to be determined.

Virtual EEG-electrodes: Convolutional neural networks as a method for upsampling or restoring channels.

BACKGROUND: In clinical practice, EEGs are assessed visually. For practical reasons, recordings often need to be performed with a reduced number of electrodes and artifacts make assessment difficult. To circumvent these obstacles, different interpolation techniques can be utilized. These techniques usually perform better for higher electrode densities and values interpolated at areas far from electrodes can be unreliable. Using a method that learns the statistical distribution of the cortical electrical fields and predicts values may yield better results. NEW METHOD: Generative networks based on convolutional layers were trained to upsample from 4 or 14 channels or to dynamically restore single missing channels to recreate 21-channel EEGs. 5,144 h of data from 1,385 subjects of the Temple University Hospital EEG database were used for training and evaluating the networks. COMPARISON WITH EXISTING METHOD: The results were compared to spherical spline interpolation. Several statistical measures were used as well as a visual evaluation by board certified clinical neurophysiologists. Overall, the generative networks performed significantly better. There was no difference between real and network generated data in the number of examples assessed as artificial by experienced EEG interpreters whereas for data generated by interpolation, the number was significantly higher. In addition, network performance improved with increasing number of included subjects, with the greatest effect seen in the range 5-100 subjects. CONCLUSIONS: Using neural networks to restore or upsample EEG signals is a viable alternative to spherical spline interpolation.

High prevalence of carpal tunnel syndrome in individuals with rare nerve growth factor-beta mutation.

In Sweden, a large family with a point mutation in the nerve growth factor-beta gene has previously been identified. The carriers of this mutation have reduced small-fibre density and selective deficits in deep pain and temperature modalities. The clinical findings in this population are described as hereditary sensory and autonomic neuropathy type V. The purpose of the current study was to investigate the prevalence of carpal tunnel syndrome in hereditary sensory and autonomic neuropathy type V based on clinical examinations and electrophysiological measurements. Furthermore, the cross-sectional area of the median nerve at the carpal tunnel inlet was measured with ultrasonography. Out of 52 known individuals heterozygous for the nerve growth factor-beta mutation in Sweden, 23 participated in the current study (12 males, 11 females; mean age 55 years; range 25-86 years). All participants answered a health questionnaire and underwent clinical examination followed by median nerve conduction study in a case-control design, and measurement of the nerve cross-sectional area with ultrasonography. The diagnosis of carpal tunnel syndrome was made based on consensus criteria using patient history and nerve conduction study. The prevalence of carpal tunnel syndrome in the hereditary sensory and autonomic neuropathy group was 35% or 52% depending on whether those individuals who had classic symptoms of carpal tunnel syndrome but negative nerve conduction studies were included or not. Those who had a high likelihood of carpal tunnel syndrome based on classic/probable patient history with positive nerve conduction study had a significantly larger median nerve cross-sectional area than those who had an unlikely patient history with negative nerve conduction study. The prevalence of carpal tunnel syndrome was 10-25 times higher in individuals heterozygous for the nerve growth factor-beta mutation than the general Swedish population. Further studies are needed to better understand the underlying pathophysiological mechanisms.

Effect of transspinal direct current stimulation on afferent pain signalling in humans.

Anodal transspinal Direct Current Stimulation (tsDCS) has been suggested as a means to treat neuropathic pain by reducing pain signalling/processing and Laser Evoked Potentials (LEPs) likewise as a method to evaluate such reduction. However, results in previous studies are disagreeing. To evaluate these claims using rigorous methodology, LEPs were evoked from hands and feet in healthy volunteers. The N2 potential and three psychophysic parameters (general- and pinprick pain, warmth) were used to evaluate the signalling and appreciation of pain respectively. This was made at three time points; at baseline, directly- and 30 min after low thoracic tsDCS (20 min, 2.5 mA, cathode on shoulder). The study was randomized, cross over, double blinded and placebo controlled. At the group level, low thoracic anodal tsDCS produced reduced perceptions of all three tested pain qualities from the foot (p < 0.05 - p < 0.001). These reductions began during stimulation and became more pronounced during the 30 min after its cessation (p < 0.05 - p < 0.01). The LEP parameter alteration mirroring these changes was latency prolongation (p < 0.05 - p < 0.001) whereas amplitude reductions were in par with placebo stimulation. Similar but less pronounced and only transient (during stimulation, p < 0.05 - p < 0.001) changes, were seen for hand stimulation. The interindividual variation was large. The findings indicate that anodal tsDCS may become a technique to treat neuropathic pain by reducing pain signalling/processing and LEPs likewise a method to evaluate such reduction.

A structural basis for sustained bacterial adhesion: biomechanical properties of CFA/I pili.

Enterotoxigenic Escherichia coli (ETEC) are a major cause of diarrheal disease worldwide. Adhesion pili (or fimbriae), such as the CFA/I (colonization factor antigen I) organelles that enable ETEC to attach efficiently to the host intestinal tract epithelium, are critical virulence factors for initiation of infection. We characterized the intrinsic biomechanical properties and kinetics of individual CFA/I pili at the single-organelle level, demonstrating that weak external forces (7.5 pN) are sufficient to unwind the intact helical filament of this prototypical ETEC pilus and that it quickly regains its original structure when the force is removed. While the general relationship between exertion of force and an increase in the filament length for CFA/I pili associated with diarrheal disease is analogous to that of P pili and type 1 pili, associated with urinary tract and other infections, the biomechanical properties of these different pili differ in key quantitative details. Unique features of CFA/I pili, including the significantly lower force required for unwinding, the higher extension speed at which the pili enter a dynamic range of unwinding, and the appearance of sudden force drops during unwinding, can be attributed to morphological features of CFA/I pili including weak layer-to-layer interactions between subunits on adjacent turns of the helix and the approximately horizontal orientation of pilin subunits with respect to the filament axis. Our results indicate that ETEC CFA/I pili are flexible organelles optimized to withstand harsh motion without breaking, resulting in continued attachment to the intestinal epithelium by the pathogenic bacteria that express these pili.