Multimodal MRI-Based Whole-Brain Assessment in Patients In Anoxoischemic Coma by Using 3D Convolutional Neural Networks.

BACKGROUND: There is an unfulfilled need to find the best way to automatically capture, analyze, organize, and merge structural and functional brain magnetic resonance imaging (MRI) data to ultimately extract relevant signals that can assist the medical decision process at the bedside of patients in postanoxic coma. We aimed to develop and validate a deep learning model to leverage multimodal 3D MRI whole-brain times series for an early evaluation of brain damages related to anoxoischemic coma. METHODS: This proof-of-concept, prospective, cohort study was undertaken at the intensive care unit affiliated with the University Hospital (Toulouse, France), between March 2018 and May 2020. All patients were scanned in coma state at least 2 days (4 ± 2 days) after cardiac arrest. Over the same period, age-matched healthy volunteers were recruited and included. Brain MRI quantification encompassed both "functional data" from regions of interest (precuneus and posterior cingulate cortex) with whole-brain functional connectivity analysis and "structural data" (gray matter volume, T1-weighted, fractional anisotropy, and mean diffusivity). A specifically designed 3D convolutional neuronal network (CNN) was created to allow conscious state discrimination (coma vs. controls) by using raw MRI indices as the input. A voxel-wise visualization method based on the study of convolutional filters was applied to support CNN outcome. The Ethics Committee of the University Teaching Hospital of Toulouse, France (2018-A31) approved the study and informed consent was obtained from all participants. RESULTS: The final cohort consisted of 29 patients in postanoxic coma and 34 healthy volunteers. Coma patients were successfully discerned from controls by using 3D CNN in combination with different MR indices. The best accuracy was achieved by functional MRI data, in particular with resting-state functional MRI of the posterior cingulate cortex, with an accuracy of 0.96 (range 0.94-0.98) on the test set from 10-time repeated tenfold cross-validation. Even more satisfactory performances were achieved through the majority voting strategy, which was able to compensate for mistakes from single MR indices. Visualization maps allowed us to identify the most relevant regions for each MRI index, notably regions previously described as possibly being involved in consciousness emergence. Interestingly, a posteriori analysis of misclassified patients indicated that they may present some common functional MRI traits with controls, which suggests further favorable outcomes. CONCLUSIONS: A fully automated identification of clinically relevant signals from complex multimodal neuroimaging data is a major research topic that may bring a radical paradigm shift in the neuroprognostication of patients with severe brain injury. We report for the first time a successful discrimination between patients in postanoxic coma patients from people serving as controls by using 3D CNN whole-brain structural and functional MRI data. Clinical Trial Number http://ClinicalTrials.gov (No. NCT03482115).

Impairment of central language processing in critically ill coronavirus disease 2019 patients with delirium.

Accumulating evidence indicates that coronavirus disease 2019 is a major cause of delirium. Given the global dimension of the current pandemic and the fact that delirium is a strong predictor of cognitive decline for critically ill patients, this raises concerns regarding the neurological cost of coronavirus disease 2019. Currently, there is a major knowledge gap related to the covert yet potentially incapacitating higher-order cognitive impairment underpinning coronavirus disease 2019 related delirium. The aim of the current study was to analyse the electrophysiological signatures of language processing in coronavirus disease 2019 patients with delirium by using a specifically designed multidimensional auditory event-related potential battery to probe hierarchical cognitive processes, including self-processing (P300) and semantic/lexical priming (N400). Clinical variables and electrophysiological data were prospectively collected in controls subjects (n = 14) and in critically ill coronavirus disease 2019 patients with (n = 19) and without (n = 22) delirium. The time from intensive care unit admission to first clinical sign of delirium was of 8 (3.5-20) days, and the delirium lasted for 7 (4.5-9.5) days. Overall, we have specifically identified in coronavirus disease 2019 patients with delirium, both a preservation of low-level central auditory processing (N100 and P200) and a coherent ensemble of covert higher-order cognitive dysfunctions encompassing self-related processing (P300) and sematic/lexical language priming (N400) (spatial-temporal clustering, P-cluster ≤ 0.05). We suggest that our results shed new light on the neuropsychological underpinnings of coronavirus disease 2019 related delirium, and may constitute a valuable method for patient's bedside diagnosis and monitoring in this clinically challenging setting.

Self-processing in coma, unresponsive wakefulness syndrome and minimally conscious state.

Introduction: Behavioral and cerebral dissociation has been now clearly established in some patients with acquired disorders of consciousness (DoC). Altogether, these studies mainly focused on the preservation of high-level cognitive markers in prolonged DoC, but did not specifically investigate lower but key-cognitive functions to consciousness emergence, such as the ability to take a first-person perspective, notably at the acute stage of coma. We made the hypothesis that the preservation of self-recognition (i) is independent of the behavioral impairment of consciousness, and (ii) can reflect the ability to recover consciousness. Methods: Hence, using bedside Electroencephalography (EEG) recordings, we acquired, in a large cohort of 129 severely brain damaged patients, the brain response to the passive listening of the subject's own name (SON) and unfamiliar other first names (OFN). One hundred and twelve of them (mean age ± SD = 46 ± 18.3 years, sex ratio M/F: 71/41) could be analyzed for the detection of an individual and significant discriminative P3 event-related brain response to the SON as compared to OFN ('SON effect', primary endpoint assessed by temporal clustering permutation tests). Results: Patients were either coma (n = 38), unresponsive wakefulness syndrome (UWS, n = 30) or minimally conscious state (MCS, n = 44), according to the revised version of the Coma Recovery Scale (CRS-R). Overall, 33 DoC patients (29%) evoked a 'SON effect'. This electrophysiological index was similar between coma (29%), MCS (23%) and UWS (34%) patients (p = 0.61). MCS patients at the time of enrolment were more likely to emerged from MCS (EMCS) at 6 months than coma and UWS patients (p = 0.013 for comparison between groups). Among the 72 survivors' patients with event-related responses recorded within 3 months after brain injury, 75% of the 16 patients with a SON effect were EMCS at 6 months, while 59% of the 56 patients without a SON effect evolved to this favorable behavioral outcome. Discussion: About 30% of severely brain-damaged patients suffering from DoC are capable to process salient self-referential auditory stimuli, even in case of absence of behavioral detection of self-conscious processing. We suggest that self-recognition covert brain ability could be an index of consciousness recovery, and thus could help to predict good outcome.