Imperfect integration: Congruency between multiple sensory sources modulates decision-making processes.

Decision-making on the basis of multiple information sources is common. However, to what extent such decisions differ from those with a single source remains unclear. We combined cognitive modelling and neural-mass modelling to characterise the neurocognitive process underlying perceptual decision-making with single or double information sources. Ninety-four human participants performed binary decisions to discriminate the coherent motion direction averaged across two independent apertures. Regardless of the angular distance of the apertures, separating motion information into two apertures resulted in a reduction in accuracy. Our cognitive and neural-mass modelling results are consistent with the hypotheses that the addition of the second information source led to a lower signal-to-noise ratio of evidence accumulation with two congruent information sources, and a change in the decision strategy of speed-accuracy trade-off with two incongruent sources. Thus, our findings support a robust behavioural change in relation to multiple information sources, which have congruency-dependent impacts on selective decision-making subcomponents.

Open Source Software for Automatic Subregional Assessment of Knee Cartilage Degradation Using Quantitative T2 Relaxometry and Deep Learning.

OBJECTIVE: We evaluated a fully automated femoral cartilage segmentation model for measuring T2 relaxation values and longitudinal changes using multi-echo spin-echo (MESE) magnetic resonance imaging (MRI). We open sourced this model and developed a web app available at https://kl.stanford.edu into which users can drag and drop images to segment them automatically. DESIGN: We trained a neural network to segment femoral cartilage from MESE MRIs. Cartilage was divided into 12 subregions along medial-lateral, superficial-deep, and anterior-central-posterior boundaries. Subregional T2 values and four-year changes were calculated using a radiologist's segmentations (Reader 1) and the model's segmentations. These were compared using 28 held-out images. A subset of 14 images were also evaluated by a second expert (Reader 2) for comparison. RESULTS: Model segmentations agreed with Reader 1 segmentations with a Dice score of 0.85 ± 0.03. The model's estimated T2 values for individual subregions agreed with those of Reader 1 with an average Spearman correlation of 0.89 and average mean absolute error (MAE) of 1.34 ms. The model's estimated four-year change in T2 for individual subregions agreed with Reader 1 with an average correlation of 0.80 and average MAE of 1.72 ms. The model agreed with Reader 1 at least as closely as Reader 2 agreed with Reader 1 in terms of Dice score (0.85 vs. 0.75) and subregional T2 values. CONCLUSIONS: Assessments of cartilage health using our fully automated segmentation model agreed with those of an expert as closely as experts agreed with one another. This has the potential to accelerate osteoarthritis research.

A computational biomarker of juvenile myoclonic epilepsy from resting-state MEG.

OBJECTIVE: For people with idiopathic generalized epilepsy, functional networks derived from their resting-state scalp electrophysiological recordings have shown an inherent higher propensity to generate seizures than those from healthy controls when assessed using the concept of brain network ictogenicity (BNI). Herein we tested whether the BNI framework is applicable to resting-state magnetoencephalography (MEG) from people with juvenile myoclonic epilepsy (JME). METHODS: The BNI framework consists in deriving a functional network from apparently normal brain activity, placing a mathematical model of ictogenicity into the network and then computing how often such network generates seizures in silico. We considered data from 26 people with JME and 26 healthy controls. RESULTS: We found that resting-state MEG functional networks from people with JME are characterized by a higher propensity to generate seizures (i.e., higher BNI) than those from healthy controls. We found a classification accuracy of 73%. CONCLUSIONS: The BNI framework is applicable to MEG and was capable of differentiating people with epilepsy from healthy controls. SIGNIFICANCE: The BNI framework may be applied to resting-state MEG to aid in epilepsy diagnosis.

Recurrence quantification analysis of dynamic brain networks.

Evidence suggests that brain network dynamics are a key determinant of brain function and dysfunction. Here we propose a new framework to assess the dynamics of brain networks based on recurrence analysis. Our framework uses recurrence plots and recurrence quantification analysis to characterize dynamic networks. For resting-state magnetoencephalographic dynamic functional networks (dFNs), we have found that functional networks recur more quickly in people with epilepsy than in healthy controls. This suggests that recurrence of dFNs may be used as a biomarker of epilepsy. For stereo electroencephalography data, we have found that dFNs involved in epileptic seizures emerge before seizure onset, and recurrence analysis allows us to detect seizures. We further observe distinct dFNs before and after seizures, which may inform neurostimulation strategies to prevent seizures. Our framework can also be used for understanding dFNs in healthy brain function and in other neurological disorders besides epilepsy.

BCIAUT-P300: A Multi-Session and Multi-Subject Benchmark Dataset on Autism for P300-Based Brain-Computer-Interfaces.

There is a lack of multi-session P300 datasets for Brain-Computer Interfaces (BCI). Publicly available datasets are usually limited by small number of participants with few BCI sessions. In this sense, the lack of large, comprehensive datasets with various individuals and multiple sessions has limited advances in the development of more effective data processing and analysis methods for BCI systems. This is particularly evident to explore the feasibility of deep learning methods that require large datasets. Here we present the BCIAUT-P300 dataset, containing 15 autism spectrum disorder individuals undergoing 7 sessions of P300-based BCI joint-attention training, for a total of 105 sessions. The dataset was used for the 2019 IFMBE Scientific Challenge organized during MEDICON 2019 where, in two phases, teams from all over the world tried to achieve the best possible object-detection accuracy based on the P300 signals. This paper presents the characteristics of the dataset and the approaches followed by the 9 finalist teams during the competition. The winner obtained an average accuracy of 92.3% with a convolutional neural network based on EEGNet. The dataset is now publicly released and stands as a benchmark for future P300-based BCI algorithms based on multiple session data.

Energy landscape of resting magnetoencephalography reveals fronto-parietal network impairments in epilepsy.

Juvenile myoclonic epilepsy (JME) is a form of idiopathic generalized epilepsy. It is yet unclear to what extent JME leads to abnormal network activation patterns. Here, we characterized statistical regularities in magnetoencephalograph (MEG) resting-state networks and their differences between JME patients and controls by combining a pairwise maximum entropy model (pMEM) and novel energy landscape analyses for MEG. First, we fitted the pMEM to the MEG oscillatory power in the front-oparietal network (FPN) and other resting-state networks, which provided a good estimation of the occurrence probability of network states. Then, we used energy values derived from the pMEM to depict an energy landscape, with a higher energy state corresponding to a lower occurrence probability. JME patients showed fewer local energy minima than controls and had elevated energy values for the FPN within the theta, beta, and gamma bands. Furthermore, simulations of the fitted pMEM showed that the proportion of time the FPN was occupied within the basins of energy minima was shortened in JME patients. These network alterations were highlighted by significant classification of individual participants employing energy values as multivariate features. Our findings suggested that JME patients had altered multistability in selective functional networks and frequency bands in the fronto-parietal cortices.

Cohort Selection for Clinical Trials From Longitudinal Patient Records: Text Mining Approach.

BACKGROUND: Clinical trials are an important step in introducing new interventions into clinical practice by generating data on their safety and efficacy. Clinical trials need to ensure that participants are similar so that the findings can be attributed to the interventions studied and not to some other factors. Therefore, each clinical trial defines eligibility criteria, which describe characteristics that must be shared by the participants. Unfortunately, the complexities of eligibility criteria may not allow them to be translated directly into readily executable database queries. Instead, they may require careful analysis of the narrative sections of medical records. Manual screening of medical records is time consuming, thus negatively affecting the timeliness of the recruitment process. OBJECTIVE: Track 1 of the 2018 National Natural Language Processing Clinical Challenge focused on the task of cohort selection for clinical trials, aiming to answer the following question: Can natural language processing be applied to narrative medical records to identify patients who meet eligibility criteria for clinical trials? The task required the participating systems to analyze longitudinal patient records to determine if the corresponding patients met the given eligibility criteria. We aimed to describe a system developed to address this task. METHODS: Our system consisted of 13 classifiers, one for each eligibility criterion. All classifiers used a bag-of-words document representation model. To prevent the loss of relevant contextual information associated with such representation, a pattern-matching approach was used to extract context-sensitive features. They were embedded back into the text as lexically distinguishable tokens, which were consequently featured in the bag-of-words representation. Supervised machine learning was chosen wherever a sufficient number of both positive and negative instances was available to learn from. A rule-based approach focusing on a small set of relevant features was chosen for the remaining criteria. RESULTS: The system was evaluated using microaveraged F measure. Overall, 4 machine algorithms, including support vector machine, logistic regression, naïve Bayesian classifier, and gradient tree boosting (GTB), were evaluated on the training data using 10-fold cross-validation. Overall, GTB demonstrated the most consistent performance. Its performance peaked when oversampling was used to balance the training data. The final evaluation was performed on previously unseen test data. On average, the F measure of 89.04% was comparable to 3 of the top ranked performances in the shared task (91.11%, 90.28%, and 90.21%). With an F measure of 88.14%, we significantly outperformed these systems (81.03%, 78.50%, and 70.81%) in identifying patients with advanced coronary artery disease. CONCLUSIONS: The holdout evaluation provides evidence that our system was able to identify eligible patients for the given clinical trial with high accuracy. Our approach demonstrates how rule-based knowledge infusion can improve the performance of machine learning algorithms even when trained on a relatively small dataset.

Breakdown of long-range temporal correlations in brain oscillations during general anesthesia.

Consciousness has been hypothesized to emerge from complex neuronal dynamics, which prevails when brain operates in a critical state. Evidence supporting this hypothesis comes mainly from studies investigating neuronal activity on a short time-scale of seconds. However, a key aspect of criticality is presence of scale-free temporal dependencies occurring across a wide range of time-scales. Indeed, robust long-range temporal correlations (LRTCs) are found in neuronal oscillations during conscious states, but it is not known how LRTCs are affected by loss of consciousness. To further test a relation between critical dynamics and consciousness, we investigated LRTCs in electrocorticography signals recorded from four macaque monkeys during resting wakefulness and general anesthesia induced by various anesthetics (ketamine, medetomidine, or propofol). Detrended Fluctuation Analysis was used to estimate LRTCs in amplitude fluctuations (envelopes) of band-pass filtered signals. We demonstrate two main findings. First, during conscious states all lateral cortical regions are characterized by significant LRTCs of alpha-band activity (7-14 Hz). LRTCs are stronger in the eyes-open than eyes-closed state, but in both states they form a spatial gradient, with anterior brain regions exhibiting stronger LRTCs than posterior regions. Second, we observed a substantial decrease of LRTCs during loss of consciousness, the magnitude of which was associated with the baseline (i.e. pre-anesthesia) state of the brain. Specifically, brain regions characterized by strongest LRTCs during a wakeful baseline exhibited greatest decreases during anesthesia (i.e. "the rich got poorer"), which consequently disturbed the posterior-anterior gradient. Therefore, our results suggest that general anesthesia affects mainly brain areas characterized by strongest LRTCs during wakefulness, which might account for lack of capacities for extensive temporal integration during loss of consciousness.