Imaging of temperature dependent hemodynamics in the rat sciatic nerve by functional photoacoustic microscopy.

BACKGROUND: Vascular hemodynamics is central to the regulation of neuro-metabolism and plays important roles in peripheral nerves diseases and their prevention. However, at present there are only a few techniques capable of directly measuring peripheral nerve vascular hemodynamics. METHOD: Here, we investigate the use of dark-field functional photoacoustic microscopy (fPAM) for intrinsic visualizing of the relative hemodynamics of the rat sciatic nerve in response to localized temperature modulation (i.e., cooling and rewarming). RESULTS AND CONCLUSION: Our main results show that the relative functional total hemoglobin concentration (HbT) is more significantly correlated with localized temperature changes than the hemoglobin oxygen saturation (SO2) changes in the sciatic nerve. Our study also indicates that the relative HbT changes are better markers of neuronal activation than SO2 during nerve temperature changes. Our results show that fPAM is a promising candidate for in vivo imaging of peripheral nerve hemodynamics without the use of contrast agents. Additionally, this technique may shed light on the neuroprotective effect of hypothermia on peripheral nerves by visualizing their intrinsic hemodynamics.

The synergy between complex channel-specific FIR filter and spatial filter for single-trial EEG classification.

The common spatial pattern analysis (CSP), a frequently utilized feature extraction method in brain-computer-interface applications, is believed to be time-invariant and sensitive to noises, mainly due to an inherent shortcoming of purely relying on spatial filtering. Therefore, temporal/spectral filtering which can be very effective to counteract the unfavorable influence of noises is usually used as a supplement. This work integrates the CSP spatial filters with complex channel-specific finite impulse response (FIR) filters in a natural and intuitive manner. Each hybrid spatial-FIR filter is of high-order, data-driven and is unique to its corresponding channel. They are derived by introducing multiple time delays and regularization into conventional CSP. The general framework of the method follows that of CSP but performs better, as proven in single-trial classification tasks like event-related potential detection and motor imagery.

Structural connectivity analysis reveals topological aberrations in patients with schizophrenia.

Topological analysis and the associated parameters allow elucidation of brain networks in health and illness. Evidently useful measures for defining network competency such as small-worldness can potentially improve understanding of brain connectivity and their disruptions underlying neuropsychiatric conditions such as schizophrenia. In the current study, we assessed the structural differences of brain networks in schizophrenia patients as compared with healthy controls. As proof of concept investigation, diffusion tensor imaging recordings from 2 schizophrenia patients and 2, gender and age matched, control subjects were subjected to analysis using several graph network distance metrics. Among them, those that appeared to have the ability to encode and highest sensitivity in shedding light about anatomical changes in neuron deficiency were the shortest path length and clustering coefficient parameters. Schizophrenia patients displayed comparatively lower clustering coefficient, longer path lengths and hence reduced small-worldness. These results suggest aberrant topological architecture in the structural brain networks of patients with schizophrenia, which may impact the psychopathological and cognitive manifestations of this potentially crippling illness.

Multiple time-lag canonical correlation analysis for removing muscular artifacts in EEG.

In this work, a new approach for joint blind source separation (BSS) of datasets at multiple time lags using canonical correlation analysis (CCA) is developed for removing muscular artifacts from electroencephalogram (EEG) recordings. The proposed approach jointly extracts sources from each dataset in a decreasing order of between-set source correlations. Muscular artifact sources that typically have lowest between-set correlations can then be removed. It is shown theoretically that the proposed use of CCA on multiple datasets at multiple time lags achieves better BSS under a more relaxed condition and hence offers better performance in removing muscular artifacts than the conventional CCA. This is further demonstrated by experiments on real EEG data.

Band specific changes in thalamocortical synchrony in field potentials after cardiac arrest induced global hypoxia.

Cardiac Arrest (CA) leads to a global hypoxic-ischemic injury in the brain leading to a poor neurological outcome. Understanding the mechanisms of functional disruption in various regions of the brain may be essential for the development of improved diagnostic and therapeutic solutions. Using controlled laboratory experiment with animal models of CA, our primary focus here is on understanding the functional changes in the thalamus and the cortex, associated with the injury and acute recovery upon resuscitation. Specifically, to study the changes in thalamocortical synchrony through these periods, we acquired local field potentials (LFPs) from the ventroposterior lateral (VPL) nucleus of the thalamus and the forelimb somatosensory cortex (S1FL) in rats after asphyxial CA. Band-specific relative Hilbert phases were used to analyze synchrony between the LFPs. We observed that the CA induced global ischemia changes the local phase-relationships by introducing a phase-lag in both the thalamus and the cortex, while the synchrony between the two regions is nearly completely lost after CA.

Frequency-domain EEG source analysis for acute tonic cold pain perception.

OBJECTIVE: To investigate electrocortical responses to tonic cold pain by frequency-domain electroencephalogram (EEG) source analysis, and to identify potential electrocortical indices of acute tonic pain. METHODS: Scalp EEG data were recorded from 26 healthy subjects under tonic cold pain (CP) and no-pain control (NP) conditions. EEG power spectra and the standardized low-resolution brain electromagnetic tomography (sLORETA) localized EEG cortical sources were compared between the two conditions in five frequency bands: 1-4 Hz, 4-8 Hz, 8-12 Hz, 12-18 Hz and 18-30 Hz. RESULTS: In line with the EEG power spectral results, the source power significantly differed between the CP and NP conditions in 8-12 Hz (CPNP) in extensive brain regions. Besides, there were also significantly different 4-8 Hz and 12-18 Hz source activities between the two conditions. Among the significant source activities, the left medial frontal and left superior frontal 4-8 Hz activities, the anterior cingulate 8-12 Hz activity and the posterior cingulate 12-18 Hz activity showed significant negative correlations with subjective pain ratings. CONCLUSIONS: The brain's perception of tonic cold pain was characterized by cortical source power changes across different frequency bands in multiple brain regions. Oscillatory activities that significantly correlated with subjective pain ratings were found in the prefrontal and cingulate regions. SIGNIFICANCE: These findings may offer useful measures for objective pain assessment and provide a basis for pain treatment by modulation of neural oscillations at specific frequencies in specific brain regions.

A spatio-temporal filtering approach to denoising of single-trial ERP in rapid image triage.

Conventional search for images containing points of interest (POI) in large-volume imagery is costly and sometimes even infeasible. The rapid image triage (RIT) system which is a human cognition guided computer vision technique is potentially a promising solution to the problem. In the RIT procedure, images are sequentially presented to a subject at a high speed. At the instant of observing a POI image, unique POI event-related potentials (ERP) characterized by P300 will be elicited and measured on the scalp. With accurate single-trial detection of such unique ERP, RIT can differentiate POI images from non-POI images. However, like other brain-computer interface systems relying on single-trial detection, RIT suffers from the low signal-to-noise ratio (SNR) of the single-trial ERP. This paper presents a spatio-temporal filtering approach tailored for the denoising of single-trial ERP for RIT. The proposed approach is essentially a non-uniformly delayed spatial Gaussian filter that attempts to suppress the non-event related background electroencephalogram (EEG) and other noises without significantly attenuating the useful ERP signals. The efficacy of the proposed approach is illustrated by both simulation tests and real RIT experiments. In particular, the real RIT experiments on 20 subjects show a statistically significant and meaningful average decrease of 9.8% in RIT classification error rate, compared to that without the proposed approach.

Common spatio-temporal pattern for single-trial detection of event-related potential in rapid serial visual presentation triage.

Searching for target images in large volume imagery is a challenging problem and the rapid serial visual presentation (RSVP) triage is potentially a promising solution to the problem. RSVP triage is essentially a cortically-coupled computer vision technique that relies on single-trial detection of event-related potentials (ERP). In RSVP triage, images are shown to a subject in a rapid serial sequence. When a target image is seen by the subject, unique ERP characterized by P300 are elicited. Thus, in RSVP triage, accurate detection of such distinct ERP allows for fast searching of target images in large volume imagery. The accuracy of the distinct ERP detection in RSVP triage depends on the feature extraction method, for which the common spatial pattern analysis (CSP) was used with limited success. This paper presents a novel feature extraction method, termed common spatio-temporal pattern (CSTP), which is critical for robust single-trial detection of ERP. Unlike the conventional CSP, whereby only spatial patterns of ERP are considered, the present proposed method exploits spatial and temporal patterns of ERP separately, providing complementary spatial and temporal features for high accurate single-trial ERP detection. Numerical study using data collected from 20 subjects in RSVP triage experiments demonstrates that the proposed method offers significant performance improvement over the conventional CSP method (corrected p-value < 0.05, Pearson r = 0.64) and other competing methods in the literature. This paper further shows that the main idea of CSTP can be easily applied to other methods.

Effect of pain perception on the heartbeat evoked potential.

OBJECTIVE: To investigate the effect of acute tonic pain on the heartbeat-evoked potential (HEP) and to test whether or not pain perception can be reflected by the HEP. METHODS: Simultaneous electroencephalogram (EEG) and electrocardiogram (ECG) were recorded from 21 healthy young adults in three conditions: passive no-task control, no-pain control and cold pain. The HEP was obtained by using ECG R-peaks as event triggers. RESULTS: Prominent HEP deflection was observed in both control conditions mainly over the frontal and central locations, while it was significantly suppressed in the cold pain condition over the right-frontal, right-central and midline locations. A comparison of the data in the first and last 5 min of cold pain condition showed that lower subjective pain ratings were accompanied by higher HEP magnitudes. A correlation analysis showed that the mean HEP magnitude over the midline locations was significantly negatively correlated with subjective pain ratings. CONCLUSIONS: Cold pain induces significant suppression of the HEP across a number of scalp locations, and the suppression is correlated with self-report of pain. SIGNIFICANCE: The HEP has the potential to serve as an alternative pain measure.

Feature selection for MLP neural network: the use of random permutation of probabilistic outputs.

This paper presents a new wrapper-based feature selection method for multilayer perceptron (MLP) neural networks. It uses a feature ranking criterion to measure the importance of a feature by computing the aggregate difference, over the feature space, of the probabilistic outputs of the MLP with and without the feature. Thus, a score of importance with respect to every feature can be provided using this criterion. Based on the numerical experiments on several artificial and real-world data sets, the proposed method performs, in general, better than several selected feature selection methods for MLP, particularly when the data set is sparse or has many redundant features. In addition, as a wrapper-based approach, the computational cost for the proposed method is modest.

Automatic EEG artifact removal: a weighted support vector machine approach with error correction.

An automatic electroencephalogram (EEG) artifact removal method is presented in this paper. Compared to past methods, it has two unique features: 1) a weighted version of support vector machine formulation that handles the inherent unbalanced nature of component classification and 2) the ability to accommodate structural information typically found in component classification. The advantages of the proposed method are demonstrated on real-life EEG recordings with comparisons made to several benchmark methods. Results show that the proposed method is preferable to the other methods in the context of artifact removal by achieving a better tradeoff between removing artifacts and preserving inherent brain activities. Qualitative evaluation of the reconstructed EEG epochs also demonstrates that after artifact removal inherent brain activities are largely preserved.

EEG-based mental fatigue measurement using multi-class support vector machines with confidence estimate.

OBJECTIVE: Automatic measurement and monitoring of mental fatigue are invaluable for preventing mental-fatigue related accidents. We test an EEG-based mental-fatigue monitoring system using a probabilistic-based support vector-machines (SVM) method. METHODS: Ten subjects underwent 25-h sleep deprivation experiments with EEG monitoring. EEG data were segmented into 3-s long epochs and manually classified into 5 mental-fatigue levels, based on subjects' performance on an auditory vigilance task (AVT). Probabilistic-based multi-class SVM and standard multi-class SVM were compared as classifiers for distinguishing mental fatigue into the 5 mental-fatigue levels. RESULTS: Accuracy of the probabilistic-based multi-class SVM was 87.2%, compared to 85.4% using the standard multi-class SVM. Using confidence estimates aggregation, accuracy increased to 91.2%. CONCLUSIONS: Probabilistic-based multi-class SVM not only gives superior classification accuracy but also provides a valuable estimate of confidence in the prediction of mental fatigue level in a given 3-s EEG epoch. SIGNIFICANCE: The work demonstrates the feasibility of an automatic EEG method for assessing and monitoring of mental fatigue. Future applications of this include traffic safety and other domains where measurement or monitoring of mental fatigue is crucial.

A feature selection method for multilevel mental fatigue EEG classification.

Two feature selection approaches for multilevel mental fatigue electroencephalogram (EEG) classification are presented in this paper, in which random forest (RF) is combined with the heuristic initial feature ranking scheme (INIT) or with the recursive feature elimination scheme (RFE). In a "leave-one-proband-out" evaluation strategy, both feature selection approaches are evaluated on the recorded mental fatigue EEG time series data from 12 subjects (each for a 25-h duration) after initial feature extractions. The latter of the two approaches performs better both in classification performance and more importantly in feature reduction. RF with RFE achieved its lowest test error rate of 12.3% using 24 top-ranked features, whereas RF with INIT reached its lowest test error rate of 15.1% using 64 top-ranked features, compared to a test error rate of 22.1% using all 304 features. The results also show that 17 key features (out of 24 top-ranked features) are consistent between the subjects using RF with RFE, which is superior to the set of 64 features as determined by RF with INIT.