Simulation of spreading depolarization trajectories in cerebral cortex: Correlation of velocity and susceptibility in patients with aneurysmal subarachnoid hemorrhage.

In many cerebral grey matter structures including the neocortex, spreading depolarization (SD) is the principal mechanism of the near-complete breakdown of the transcellular ion gradients with abrupt water influx into neurons. Accordingly, SDs are abundantly recorded in patients with traumatic brain injury, spontaneous intracerebral hemorrhage, aneurysmal subarachnoid hemorrhage (aSAH) and malignant hemispheric stroke using subdural electrode strips. SD is observed as a large slow potential change, spreading in the cortex at velocities between 2 and 9 mm/min. Velocity and SD susceptibility typically correlate positively in various animal models. In patients monitored in neurocritical care, the Co-Operative Studies on Brain Injury Depolarizations (COSBID) recommends several variables to quantify SD occurrence and susceptibility, although accurate measures of SD velocity have not been possible. Therefore, we developed an algorithm to estimate SD velocities based on reconstructing SD trajectories of the wave-front's curvature center from magnetic resonance imaging scans and time-of-SD-arrival-differences between subdural electrode pairs. We then correlated variables indicating SD susceptibility with algorithm-estimated SD velocities in twelve aSAH patients. Highly significant correlations supported the algorithm's validity. The trajectory search failed significantly more often for SDs recorded directly over emerging focal brain lesions suggesting in humans similar to animals that the complexity of SD propagation paths increase in tissue undergoing injury.

Challenges in representation learning: a report on three machine learning contests.

The ICML 2013 Workshop on Challenges in Representation Learning(1) focused on three challenges: the black box learning challenge, the facial expression recognition challenge, and the multimodal learning challenge. We describe the datasets created for these challenges and summarize the results of the competitions. We provide suggestions for organizers of future challenges and some comments on what kind of knowledge can be gained from machine learning competitions.

Correlates of spreading depolarization in human scalp electroencephalography.

It has been known for decades that suppression of spontaneous scalp electroencephalographic activity occurs during ischaemia. Trend analysis for such suppression was found useful for intraoperative monitoring during carotid endarterectomy, or as a screening tool to detect delayed cerebral ischaemia after aneurismal subarachnoid haemorrhage. Nevertheless, pathogenesis of such suppression of activity has remained unclear. In five patients with aneurismal subarachnoid haemorrhage and four patients with decompressive hemicraniectomy after malignant hemispheric stroke due to middle cerebral artery occlusion, we here performed simultaneously full-band direct and alternating current electroencephalography at the scalp and direct and alternating current electrocorticography at the cortical surface. After subarachnoid haemorrhage, 275 slow potential changes, identifying spreading depolarizations, were recorded electrocorticographically over 694 h. Visual inspection of time-compressed scalp electroencephalography identified 193 (70.2%) slow potential changes [amplitude: -272 (-174, -375) µV (median quartiles), duration: 5.4 (4.0, 7.1) min, electrocorticography-electroencephalography delay: 1.8 (0.8, 3.5) min]. Intervals between successive spreading depolarizations were significantly shorter for depolarizations with electroencephalographically identified slow potential change [33.0 (27.0, 76.5) versus 53.0 (28.0, 130.5) min, P = 0.009]. Electroencephalography was thus more likely to display slow potential changes of clustered than isolated spreading depolarizations. In contrast to electrocorticography, no spread of electroencephalographic slow potential changes was seen, presumably due to superposition of volume-conducted electroencephalographic signals from widespread cortical generators. In two of five patients with subarachnoid haemorrhage, serial magnetic resonance imaging revealed large delayed infarcts at the recording site, while electrocorticography showed clusters of spreading depolarizations with persistent depression of spontaneous activity. Alternating current electroencephalography similarly displayed persistent depression of spontaneous activity, and direct current electroencephalography slow potential changes riding on a shallow negative ultraslow potential. Isolated spreading depolarizations with depression of both spontaneous electrocorticographic and electroencephalographic activity displayed significantly longer intervals between successive spreading depolarizations than isolated depolarizations with only depression of electrocorticographic activity [44.0 (28.0, 132.0) min, n = 96, versus 30.0 (26.5, 51.5) min, n = 109, P = 0.001]. This suggests fusion of electroencephalographic depression periods at high depolarization frequency. No propagation of electroencephalographic depression was seen between scalp electrodes. Durations/magnitudes of isolated electroencephalographic and corresponding electrocorticographic depression periods correlated significantly. Fewer spreading depolarizations were recorded in patients with malignant hemispheric stroke but characteristics were similar to those after subarachnoid haemorrhage. In conclusion, spreading depolarizations and depressions of spontaneous activity display correlates in time-compressed human scalp direct and alternating current electroencephalography that may serve for their non-invasive detection.

Bristle-sensors--low-cost flexible passive dry EEG electrodes for neurofeedback and BCI applications.

In this paper, we present a new, low-cost dry electrode for EEG that is made of flexible metal-coated polymer bristles. We examine various standard EEG paradigms, such as capturing occipital alpha rhythms, testing for event-related potentials in an auditory oddball paradigm and performing a sensory motor rhythm-based event-related (de-) synchronization paradigm to validate the performance of the novel electrodes in terms of signal quality. Our findings suggest that the dry electrodes that we developed result in high-quality EEG recordings and are thus suitable for a wide range of EEG studies and BCI applications. Furthermore, due to the flexibility of the novel electrodes, greater comfort is achieved in some subjects, this being essential for long-term use.

Single-subject classification of schizophrenia by event-related potentials during selective attention.

BACKGROUND: Executive dysfunction has repeatedly been proposed as a robust and promising substrate of analytical approaches in the research of neurocognitive markers of schizophrenia. Here, we present a mixed model- and data-driven classification approach by applying a task that targets executive dysfunction in schizophrenia and by investigating relevant event-related potential (ERP) features with machine learning classifiers. METHODS: Forty schizophrenic patients and forty matched healthy controls completed the Attention Network Test while an electroencephalogram was recorded. Target-locked N1 and P3 ERP components were constructed and submitted to different classification analyses without a priori hypotheses. Standardized source localization was applied to estimate neural sources of N1 and P3 deficits in schizophrenia. RESULTS: We obtained a classification accuracy of 79% using only very few ERP components. Central P3 components following compatible and incompatible trials and right parietal N1 latencies averaged across targets and were sufficient for classification. P3 deficits were associated with anterior cingulate cortex dysfunction, while right posterior current density deficits were observed in schizophrenia during the N1 time frame. CONCLUSIONS: The data exemplarily show how automated inference may be applied to classify a pathological state in single subjects without prior knowledge of their diagnoses. While classification accuracy may be optimized by application of other executive paradigms, this approach illustrates the potential of machine learning algorithms for the identification of biomarkers that are independent of clinical assessments. Conversely, data suggest a pathophysiological mechanism that includes early visual and late executive deficits during response inhibition in schizophrenia.

Subject-independent mental state classification in single trials.

Current state-of-the-art in Brain Computer Interfacing (BCI) involves tuning classifiers to subject-specific training data acquired from calibration sessions prior to functional BCI use. Using a large database of EEG recordings from 45 subjects, who took part in movement imagination task experiments, we construct an ensemble of classifiers derived from subject-specific temporal and spatial filters. The ensemble is then sparsified using quadratic regression with l(1) regularization such that the final classifier generalizes reliably to data of subjects not included in the ensemble. Our offline results indicate that BCI-naïve users could start real-time BCI use without any prior calibration at only very limited loss of performance.