Task-free spectral EEG dynamics track and predict patient recovery from severe acquired brain injury

A Long Short-Term Memory Network Using Resting-State Electroencephalogram to Predict Outcomes Following Moderate Traumatic Brain Injury

Although traumatic brain injury (TBI) is a global public health issue, not all injuries necessitate additional hospitalisation. Thinking, memory, attention, personality, and movement can all be negatively impacted by TBI. However, only a small proportion of nonsevere TBIs necessitate prolonged observation. Clinicians would benefit from an electroencephalography (EEG)-based computational intelligence model for outcome prediction by having access to an evidence-based analysis that would allow them to securely discharge patients who are at minimal risk of TBI-related mortality. Despite the increasing popularity of EEG-based deep learning research to create predictive models with breakthrough performance, particularly in epilepsy prediction, its use in clinical decision making for the diagnosis and prognosis of TBI has not been as widely exploited. Therefore, utilising 60s segments of unprocessed resting-state EEG data as input, we suggest a long short-term memory (LSTM) network that can distinguish between improved and unimproved outcomes in moderate TBI patients. Complex feature extraction and selection are avoided in this architecture. The experimental results show that, with a classification accuracy of 87.50 ± 0.05%, the proposed prognostic model outperforms three related works. The results suggest that the proposed methodology is an efficient and reliable strategy to assist clinicians in creating an automated tool for predicting treatment outcomes from EEG signals.

Brain Activity Characteristics of Patients With Disorders of Consciousness in the EEG Resting State Paradigm: A Review

The assessment of the level of consciousness in disorders of consciousness (DoC) is still one of the most challenging problems in contemporary medicine. Nevertheless, based on the multitude of studies conducted over the last 20 years on resting states based on electroencephalography (EEG) in DoC, it is possible to outline the brain activity profiles related to both patients without preserved consciousness and minimally conscious ones. In the case of patients without preserved consciousness, the dominance of low, mostly delta, frequency, and the marginalization of the higher frequencies were observed, both in terms of the global power of brain activity and in functional connectivity patterns. In turn, the minimally conscious patients revealed the opposite brain activity pattern—the characteristics of higher frequency bands were preserved both in global power and in functional long-distance connections. In this short review, we summarize the state of the art of EEG-based research in the resting state paradigm, in the context of providing potential support to the traditional clinical assessment of the level of consciousness.

Brain Data in Pediatric Disorders of Consciousness: Special Considerations

SUMMARY

The diagnosis and management of disorders of consciousness in children continue to present a clinical, research, and ethical challenge. Though the practice guidelines for diagnosis and management of disorders of consciousness in adults are supported by decades of empirical and pragmatic evidence, similar guidelines for infants and children are lacking. The maturing conscious experience and the limited behavioral repertoire to report consciousness in this age group restrict extrapolation from the adult literature. Equally challenging is the process of heightened structural and functional neuroplasticity in the developing brain, which adds a layer of complexity to the investigation of the neural correlates of consciousness in infants and children. This review discusses the clinical assessment of pediatric disorders of consciousness and delineates the diagnostic and prognostic utility of neurophysiological and neuroimaging correlates of consciousness. The potential relevance of these correlates for the developing brain based on existing theoretical models of consciousness in adults is outlined.

Spatio-temporal analysis of EEG features during consciousness recovery in patients with disorders of consciousness

OBJECTIVE

As consciousness recovery is not only dynamic but also involves interactions between various brain regions, elucidating the mechanism of recovery requires tracking cortical activity in spatio-temporal dimensions.

METHODS

We tracked the cortical activities of 40 patients (mean age: 54.38 years; 28 males; 21 patients with minimally conscious states) with disorders of consciousness, and collected a total of 156 electroencephalographic signals. We investigated the longitudinal changes in EEG nonlinear dynamic features (i.e., approximate entropy, sample entropy, and Lempel-Ziv complexity) and relative wavelet energy along with consciousness recovery.

RESULTS

Global EEG features showed a non-monotonic trend during consciousness recovery (P < 0.05). When the level of consciousness of patients was transferred to a minimally conscious state from an unresponsive wakefulness syndrome/ vegetative state, an inflection point appeared in the EEG features. The EEG feature change trends between the injured and uninjured areas were dissimilar (P < 0.05). Importantly, the degree of dissimilarity increased non-monotonically across the levels of consciousness (P < 0.05).

CONCLUSIONS

EEG recovery was non-monotonic and dissimilar in spatio-temporal dimensions, with an inflection point.

SIGNIFICANCE

These findings further clarify the process of consciousness recovery and provide assistance in exploring the mechanism of consciousness recovery.

Biomarkers for Traumatic Brain Injury: Data Standards and Statistical Considerations

Recent biomarker innovations hold potential for transforming diagnosis, prognostic modeling, and precision therapeutic targeting of traumatic brain injury (TBI). However, many biomarkers, including brain imaging, genomics, and proteomics, involve vast quantities of high-throughput and high-content data. Management, curation, analysis, and evidence synthesis of these data are not trivial tasks. In this review, we discuss data management concepts and statistical and data sharing strategies when dealing with biomarker data in the context of TBI research. We propose that application of biomarkers involves three distinct steps-discovery, evaluation, and evidence synthesis. First, complex/big data has to be reduced to useful data elements at the stage of biomarker discovery. Second, inferential statistical approaches must be applied to these biomarker data elements for assessment of biomarker clinical utility and validity. Last, synthesis of relevant research is required to support practice guidelines and enable health decisions informed by the highest quality, up-to-date evidence available. We focus our discussion around recent experiences from the International Traumatic Brain Injury Research (InTBIR) initiative, with a specific focus on four major clinical projects (Transforming Research and Clinical Knowledge in TBI, Collaborative European NeuroTrauma Effectiveness Research in TBI, Collaborative Research on Acute Traumatic Brain Injury in Intensive Care Medicine in Europe, and Approaches and Decisions in Acute Pediatric TBI Trial), which are currently enrolling subjects in North America and Europe. We discuss common data elements, data collection efforts, data-sharing opportunities, and challenges, as well as examine the statistical techniques required to realize successful adoption and use of biomarkers in the clinic as a foundation for precision medicine in TBI.

Current uses, emerging applications, and clinical integration of artificial intelligence in neuroradiology

Artificial intelligence (AI) is a branch of computer science with a variety of subfields and techniques, exploited to serve as a deductive tool that performs tasks originally requiring human cognition. AI tools and its subdomains are being incorporated into healthcare delivery for the improvement of medical data interpretation encompassing clinical management, diagnostics, and prognostic outcomes. In the field of neuroradiology, AI manifested through deep machine learning and connected neural networks (CNNs) has demonstrated incredible accuracy in identifying pathology and aiding in diagnosis and prognostication in several areas of neurology and neurosurgery. In this literature review, we survey the available clinical data highlighting the utilization of AI in the field of neuroradiology across multiple neurological and neurosurgical subspecialties. In addition, we discuss the emerging role of AI in neuroradiology, its strengths and limitations, as well as future needs in strengthening its role in clinical practice. Our review evaluated data across several subspecialties of neurology and neurosurgery including vascular neurology, spinal pathology, traumatic brain injury (TBI), neuro-oncology, multiple sclerosis, Alzheimer's disease, and epilepsy. AI has established a strong presence within the realm of neuroradiology as a successful and largely supportive technology aiding in the interpretation, diagnosis, and even prognostication of various pathologies. More research is warranted to establish its full scientific validity and determine its maximum potential to aid in optimizing and providing the most accurate imaging interpretation.

Electrophysiological and Neuroimaging Studies - During Resting State and Sensory Stimulation in Disorders of Consciousness: A Review

A severe brain injury may lead to a disorder of consciousness (DOC) such as coma, vegetative state (VS), minimally conscious state (MCS) or locked-in syndrome (LIS). Till date, the diagnosis of DOC relies only on clinical evaluation or subjective scoring systems such as Glasgow coma scale, which fails to detect subtle changes and thereby results in diagnostic errors. The high rate of misdiagnosis and inability to predict the recovery of consciousness for DOC patients have created a huge research interest in the assessment of consciousness. Researchers have explored the use of various stimulation and neuroimaging techniques to improve the diagnosis. In this article, we present the important findings of resting-state as well as sensory stimulation methods and highlight the stimuli proven to be successful in the assessment of consciousness. Primarily, we review the literature based on (a) application/non-use of stimuli (i.e., sensory stimulation/resting state-based), (b) type of stimulation used (i.e., auditory, visual, tactile, olfactory, or mental-imagery), (c) electrophysiological signal used (EEG/ERP, fMRI, PET, EMG, SCL, or ECG). Among the sensory stimulation methods, auditory stimulation has been extensively used, since it is easier to conduct for these patients. Olfactory and tactile stimulation have been less explored and need further research. Emotionally charged stimuli such as subject’s own name or narratives in a familiar voice or subject’s own face/family pictures or music result in stronger responses than neutral stimuli. Studies based on resting state analysis have employed measures like complexity, power spectral features, entropy and functional connectivity patterns to distinguish between the VS and MCS patients. Resting-state EEG and fMRI are the state-of-the-art techniques and have a huge potential in predicting the recovery of coma patients. Further, EMG and mental-imagery based studies attempt to obtain volitional responses from the VS patients and thus could detect their command-following capability. This may provide an effective means to communicate with these patients. Recent studies have employed fMRI and PET to understand the brain-activation patterns corresponding to the mental imagery. This review promotes our knowledge about the techniques used for the diagnosis of patients with DOC and attempts to provide ideas for future research.

Classification of Non-Severe Traumatic Brain Injury from Resting-State EEG Signal Using LSTM Network with ECOC-SVM

Traumatic brain injury (TBI) is one of the common injuries when the human head receives an impact due to an accident or fall and is one of the most frequently submitted insurance claims. However, it is often always misused when individuals attempt an insurance fraud claim by providing false medical conditions. Therefore, there is a need for an instant brain condition classification system. This study presents a novel classification architecture that can classify non-severe TBI patients and healthy subjects employing resting-state electroencephalogram (EEG) as the input, solving the immobility issue of the computed tomography (CT) scan and magnetic resonance imaging (MRI). The proposed architecture makes use of long short term memory (LSTM) and error-correcting output coding support vector machine (ECOC-SVM) to perform multiclass classification. The pre-processed EEG time series are supplied to the network by each time step, where important information from the previous time step will be remembered by the LSTM cell. Activations from the LSTM cell is used to train an ECOC-SVM. The temporal advantages of the EEG were amplified and able to achieve a classification accuracy of 100%. The proposed method was compared to existing works in the literature, and it is shown that the proposed method is superior in terms of classification accuracy, sensitivity, specificity, and precision.

Hamid AI, Abdullah MZ, Azman A, Abdullah JM. Detection of Moderate Traumatic Brain Injury from Resting-State Eye-Closed Electroencephalography

Traumatic brain injury (TBI) is one of the injuries that can bring serious consequences if medical attention has been delayed. Commonly, analysis of computed tomography (CT) or magnetic resonance imaging (MRI) is required to determine the severity of a moderate TBI patient. However, due to the rising number of TBI patients these days, employing the CT scan or MRI scan to every potential patient is not only expensive, but also time consuming. Therefore, in this paper, we investigate the possibility of using electroencephalography (EEG) with computational intelligence as an alternative approach to detect the severity of moderate TBI patients. EEG procedure is much cheaper than CT or MRI. Although EEG does not have high spatial resolutions as compared with CT and MRI, it has high temporal resolutions. The analysis and prediction of moderate TBI from EEG using conventional computational intelligence approaches are tedious as they normally involve complex preprocessing, feature extraction, or feature selection of the signal. Thus, we propose an approach that uses convolutional neural network (CNN) to automatically classify healthy subjects and moderate TBI patients. The input to this computational intelligence system is the resting-state eye-closed EEG, without undergoing preprocessing and feature selection. The EEG dataset used includes 15 healthy volunteers and 15 moderate TBI patients, which is acquired at the Hospital Universiti Sains Malaysia, Kelantan, Malaysia. The performance of the proposed method has been compared with four other existing methods. With the average classification accuracy of 72.46%, the proposed method outperforms the other four methods. This result indicates that the proposed method has the potential to be used as a preliminary screening for moderate TBI, for selection of the patients for further diagnosis and treatment planning.

Small-World Propensity Reveals the Frequency Specificity of Resting State Networks

Goal: Functional connectivity (FC) is an important indicator of the brain's state in different conditions, such as rest/task or health/pathology. Here we used high-density electroencephalography coupled to source reconstruction to assess frequency-specific changes of FC during resting state. Specifically, we computed the Small-World Propensity (SWP) index to characterize network small-world architecture across frequencies. Methods: We collected resting state data from healthy participants and built connectivity matrices maintaining the heterogeneity of connection strengths. For a subsample of participants, we also investigated whether the SWP captured FC changes after the execution of a working memory (WM) task. Results: We found that SWP demonstrated a selective increase in the alpha and low beta bands. Moreover, SWP was modulated by a cognitive task and showed increased values in the bands entrained by the WM task. Conclusions: SWP is a valid metric to characterize the frequency-specific behavior of resting state networks.

Pharmacologically informed machine learning approach for identifying pathological states of unconsciousness via resting-state fMRI

Determining the level of consciousness in patients with disorders of consciousness (DOC) remains challenging. To address this challenge, resting-state fMRI (rs-fMRI) has been widely used for detecting the local, regional, and network activity differences between DOC patients and healthy controls. Although substantial progress has been made towards this endeavor, the identification of robust rs-fMRI-based biomarkers for level of consciousness is still lacking. Recent developments in machine learning show promise as a tool to augment the discrimination between different states of consciousness in clinical practice. Here, we investigated whether machine learning models trained to make a binary distinction between conscious wakefulness and anesthetic-induced unconsciousness would then be capable of reliably identifying pathologically induced unconsciousness. We did so by extracting rs-fMRI-based features associated with local activity, regional homogeneity, and interregional functional activity in 44 subjects during wakefulness, light sedation, and unresponsiveness (deep sedation and general anesthesia), and subsequently using those features to train three distinct candidate machine learning classifiers: support vector machine, Extra Trees, artificial neural network. First, we show that all three classifiers achieve reliable performance within-dataset (via nested cross-validation), with a mean area under the receiver operating characteristic curve (AUC) of 0.95, 0.92, and 0.94, respectively. Additionally, we observed comparable cross-dataset performance (making predictions on the DOC data) as the anesthesia-trained classifiers demonstrated a consistent ability to discriminate between unresponsive wakefulness syndrome (UWS/VS) patients and healthy controls with mean AUC's of 0.99, 0.94, 0.98, respectively. Lastly, we explored the potential of applying the aforementioned classifiers towards discriminating intermediate states of consciousness, specifically, subjects under light anesthetic sedation and patients diagnosed as having a minimally conscious state (MCS). Our findings demonstrate that machine learning classifiers trained on rs-fMRI features derived from participants under anesthesia have potential to aid the discrimination between degrees of pathological unconsciousness in clinical patients.

Brainstem Modulation of Large-Scale Intrinsic Cortical Activity Correlations

Brain activity fluctuates continuously, even in the absence of changes in sensory input or motor output. These intrinsic activity fluctuations are correlated across brain regions and are spatially organized in macroscale networks. Variations in the strength, topography, and topology of correlated activity occur over time, and unfold upon a backbone of long-range anatomical connections. Subcortical neuromodulatory systems send widespread ascending projections to the cortex, and are thus ideally situated to shape the temporal and spatial structure of intrinsic correlations. These systems are also the targets of the pharmacological treatment of major neurological and psychiatric disorders, such as Parkinson's disease, depression, and schizophrenia. Here, we review recent work that has investigated how neuromodulatory systems shape correlations of intrinsic fluctuations of large-scale cortical activity. We discuss studies in the human, monkey, and rodent brain, with a focus on non-invasive recordings of human brain activity. We provide a structured but selective overview of this work and distil a number of emerging principles. Future efforts to chart the effect of specific neuromodulators and, in particular, specific receptors, on intrinsic correlations may help identify shared or antagonistic principles between different neuromodulatory systems. Such principles can inform models of healthy brain function and may provide an important reference for understanding altered cortical dynamics that are evident in neurological and psychiatric disorders, potentially paving the way for mechanistically inspired biomarkers and individualized treatments of these disorders.

Challenges and demand for modeling disorders of consciousness following traumatic brain injury

Following severe traumatic brain injury (TBI), many patients experience coma - an unresponsive state lacking wakefulness or awareness. Coma rarely lasts more than two weeks, and emergence involves passing through a state of wakefulness without awareness of self or environment. Patients that linger in these Disorders of Consciousness (DoC) undergo clinical assessments of awareness for diagnosis into Unresponsive Wakefulness Syndrome (no awareness, also called vegetative state) or Minimally Conscious State (periodic increases in awareness). These diagnoses are notoriously inaccurate, offering little prognostic value. Recovery of awareness is unpredictable, returning within weeks, years, or never. This leaves patients' families with difficult decisions and little information on which to base them. Clinical studies have made significant advancements, but remain encumbered by high variability, limited data output, and a lack of necessary controls. Herein we discuss the clear and present need to establish a preclinical model of TBI-induced DoC, the significant challenges involved, and how such a model can be applied to support DoC research.