Decoding P300 Variability Using Convolutional Neural Networks

Saccade size predicts onset time of object processing during visual search of an open world virtual environment

OBJECTIVE

To date the vast majority of research in the visual neurosciences have been forced to adopt a highly constrained perspective of the vision system in which stimuli are processed in an open-loop reactive fashion (i.e., abrupt stimulus presentation followed by an evoked neural response). While such constraints enable high construct validity for neuroscientific investigation, the primary outcomes have been a reductionistic approach to isolate the component processes of visual perception. In electrophysiology, of the many neural processes studied under this rubric, the most well-known is, arguably, the P300 evoked response. There is, however, relatively little known about the real-world corollary of this component in free-viewing paradigms where visual stimuli are connected to neural function in a closed-loop. While growing evidence suggests that neural activity analogous to the P300 does occur in such paradigms, it is an open question when this response occurs and what behavioral or environmental factors could be used to isolate this component.

APPROACH

The current work uses convolutional networks to decode neural signals during a free-viewing visual search task in a closed-loop paradigm within an open-world virtual environment. From the decoded activity we construct fixation-locked response profiles that enable estimations of the variable latency of any P300 analogue around the moment of fixation. We then use these estimates to investigate which factors best reduce variable latency and, thus, predict the onset time of the response. We consider measurable, search-related factors encompassing top-down (i.e., goal driven) and bottom-up (i.e., stimulus driven) processes, such as fixation duration and salience. We also consider saccade size as an intermediate factor reflecting the integration of these two systems.

MAIN RESULTS

The results show that of these factors only saccade size reliably determines the onset time of P300 analogous activity for this task. Specifically, we find that for large saccades the variability in response onset is small enough to enable analysis using traditional ensemble averaging methods.

SIGNIFICANCE

The results show that P300 analogous activity does occur during closed-loop, free-viewing visual search while highlighting distinct differences between the open-loop version of this response and its real-world analogue. The results also further establish saccades, and saccade size, as a key factor in real-world visual processing.

Evidence of elevated situational awareness for active duty soldiers during navigation of a virtual environment

U.S. service members maintain constant situational awareness (SA) due to training and experience operating in dynamic and complex environments. Work examining how military experience impacts SA during visual search of a complex naturalistic environment, is limited. Here, we compare Active Duty service members and Civilians' physiological behavior during a navigational visual search task in an open-world virtual environment (VE) while cognitive load was manipulated. We measured eye-tracking and electroencephalogram (EEG) outcomes from Active Duty (N = 21) and Civilians (N = 15) while they navigated a desktop VE at a self-regulated pace. Participants searched and counted targets (N = 15) presented among distractors, while cognitive load was manipulated with an auditory Math Task. Results showed Active Duty participants reported significantly greater/closer to the correct number of targets compared to Civilians. Overall, Active Duty participants scanned the VE with faster peak saccade velocities and greater average saccade magnitudes compared to Civilians. Convolutional Neural Network (CNN) response (EEG P-300) was significantly weighted more to initial fixations for the Active Duty group, showing reduced attentional resources on object refixations compared to Civilians. There were no group differences in fixation outcomes or overall CNN response when comparing targets versus distractor objects. When cognitive load was manipulated, only Civilians significantly decreased their average dwell time on each object and the Active Duty group had significantly fewer numbers of correct answers on the Math Task. Overall, the Active Duty group explored the VE with increased scanning speed and distance and reduced cognitive re-processing on objects, employing a different, perhaps expert, visual search strategy indicative of increased SA. The Active Duty group maintained SA in the main visual search task and did not appear to shift focus to the secondary Math Task. Future work could compare how a stress inducing environment impacts these groups' physiological or cognitive markers and performance for these groups.

Decoding neural activity to assess individual latent state in ecologically valid contexts

Objective.Currently, there exists very few ways to isolate cognitive processes, historically defined via highly controlled laboratory studies, in more ecologically valid contexts. Specifically, it remains unclear as to what extent patterns of neural activity observed under such constraints actually manifest outside the laboratory in a manner that can be used to make accurate inferences about latent states, associated cognitive processes, or proximal behavior. Improving our understanding of when and how specific patterns of neural activity manifest in ecologically valid scenarios would provide validation for laboratory-based approaches that study similar neural phenomena in isolation and meaningful insight into the latent states that occur during complex tasks.Approach.Domain generalization methods, borrowed from the work of the brain-computer interface community, have the potential to capture high-dimensional patterns of neural activity in a way that can be reliably applied across experimental datasets in order to address this specific challenge. We previously used such an approach to decode phasic neural responses associated with visual target discrimination. Here, we extend that work to more tonic phenomena such as internal latent states. We use data from two highly controlled laboratory paradigms to train two separate domain-generalized models. We apply the trained models to an ecologically valid paradigm in which participants performed multiple, concurrent driving-related tasks while perched atop a six-degrees-of-freedom ride-motion simulator.Main Results.Using the pretrained models, we estimate latent state and the associated patterns of neural activity. As the patterns of neural activity become more similar to those patterns observed in the training data, we find changes in behavior and task performance that are consistent with the observations from the original, laboratory-based paradigms.Significance.These results lend ecological validity to the original, highly controlled, experimental designs and provide a methodology for understanding the relationship between neural activity and behavior during complex tasks.

Deep learning in neuroimaging data analysis: Applications, challenges, and solutions

Methods for the analysis of neuroimaging data have advanced significantly since the beginning of neuroscience as a scientific discipline. Today, sophisticated statistical procedures allow us to examine complex multivariate patterns, however most of them are still constrained by assuming inherent linearity of neural processes. Here, we discuss a group of machine learning methods, called deep learning, which have drawn much attention in and outside the field of neuroscience in recent years and hold the potential to surpass the mentioned limitations. Firstly, we describe and explain the essential concepts in deep learning: the structure and the computational operations that allow deep models to learn. After that, we move to the most common applications of deep learning in neuroimaging data analysis: prediction of outcome, interpretation of internal representations, generation of synthetic data and segmentation. In the next section we present issues that deep learning poses, which concerns multidimensionality and multimodality of data, overfitting and computational cost, and propose possible solutions. Lastly, we discuss the current reach of DL usage in all the common applications in neuroimaging data analysis, where we consider the promise of multimodality, capability of processing raw data, and advanced visualization strategies. We identify research gaps, such as focusing on a limited number of criterion variables and the lack of a well-defined strategy for choosing architecture and hyperparameters. Furthermore, we talk about the possibility of conducting research with constructs that have been ignored so far or/and moving toward frameworks, such as RDoC, the potential of transfer learning and generation of synthetic data.

Age-related differences in visual P300 ERP during dual-task postural balance

Standing and concurrently performing a cognitive task is a very common situation in everyday life. It is associated with a higher risk of falling in the elderly. Here, we aim at evaluating the differences of the P300 evoked potential elicited by a visual oddball paradigm between healthy younger (< 35 y) and older (> 64 y) adults during a simultaneous postural task. We found that P300 latency increases significantly (p < 0.001) when the elderly are engaged in more challenging postural tasks; younger adults show no effect of balance condition. Our results demonstrate that, even if the elderly have the same accuracy in odd stimuli detection as younger adults do, they require a longer processing time for stimulus discrimination. This finding suggests an increased attentional load which engages additional cerebral reserves.

A Lightweight Multi-Scale Convolutional Neural Network for P300 Decoding: Analysis of Training Strategies and Uncovering of Network Decision

Convolutional neural networks (CNNs), which automatically learn features from raw data to approximate functions, are being increasingly applied to the end-to-end analysis of electroencephalographic (EEG) signals, especially for decoding brain states in brain-computer interfaces (BCIs). Nevertheless, CNNs introduce a large number of trainable parameters, may require long training times, and lack in interpretability of learned features. The aim of this study is to propose a CNN design for P300 decoding with emphasis on its lightweight design while guaranteeing high performance, on the effects of different training strategies, and on the use of post-hoc techniques to explain network decisions. The proposed design, named MS-EEGNet, learned temporal features in two different timescales (i.e., multi-scale, MS) in an efficient and optimized (in terms of trainable parameters) way, and was validated on three P300 datasets. The CNN was trained using different strategies (within-participant and within-session, within-participant and cross-session, leave-one-subject-out, transfer learning) and was compared with several state-of-the-art (SOA) algorithms. Furthermore, variants of the baseline MS-EEGNet were analyzed to evaluate the impact of different hyper-parameters on performance. Lastly, saliency maps were used to derive representations of the relevant spatio-temporal features that drove CNN decisions. MS-EEGNet was the lightest CNN compared with the tested SOA CNNs, despite its multiple timescales, and significantly outperformed the SOA algorithms. Post-hoc hyper-parameter analysis confirmed the benefits of the innovative aspects of MS-EEGNet. Furthermore, MS-EEGNet did benefit from transfer learning, especially using a low number of training examples, suggesting that the proposed approach could be used in BCIs to accurately decode the P300 event while reducing calibration times. Representations derived from the saliency maps matched the P300 spatio-temporal distribution, further validating the proposed decoding approach. This study, by specifically addressing the aspects of lightweight design, transfer learning, and interpretability, can contribute to advance the development of deep learning algorithms for P300-based BCIs.

An Introductory Tutorial on Brain–Computer Interfaces and Their Applications

The prospect and potentiality of interfacing minds with machines has long captured human imagination. Recent advances in biomedical engineering, computer science, and neuroscience are making brain–computer interfaces a reality, paving the way to restoring and potentially augmenting human physical and mental capabilities. Applications of brain–computer interfaces are being explored in applications as diverse as security, lie detection, alertness monitoring, gaming, education, art, and human cognition augmentation. The present tutorial aims to survey the principal features and challenges of brain–computer interfaces (such as reliable acquisition of brain signals, filtering and processing of the acquired brainwaves, ethical and legal issues related to brain–computer interface (BCI), data privacy, and performance assessment) with special emphasis to biomedical engineering and automation engineering applications. The content of this paper is aimed at students, researchers, and practitioners to glimpse the multifaceted world of brain–computer interfacing.

Toward Measuring Target Perception: First-Order and Second-Order Deep Network Pipeline for Classification of Fixation-Related Potentials

The topdown determined visual object perception refers to the ability of a person to identify a prespecified visual target. This paper studies the technical foundation for measuring the target-perceptual ability in a guided visual search task, using the EEG-based brain imaging technique. Specifically, it focuses on the feature representation learning problem for single-trial classification of fixation-related potentials (FRPs). The existing methods either capture only first-order statistics while ignoring second-order statistics in data, or directly extract second-order statistics with covariance matrices estimated with raw FRPs that suffer from low signal-to-noise ratio. In this paper, we propose a new representation learning pipeline involving a low-level convolution subnetwork followed by a high-level Riemannian manifold subnetwork, with a novel midlevel pooling layer bridging them. In this way, the discriminative power of the first-order features can be increased by the convolution subnetwork, while the second-order information in the convolutional features could further be deeply learned with the subsequent Riemannian subnetwork. In particular, the temporal ordering of FRPs is well preserved for the components in our pipeline, which is considered to be a valuable source of discriminant information. The experimental results show that proposed approach leads to improved classification performance and robustness to lack of data over the state-of-the-art ones, thus making it appealing for practical applications in measuring the target-perceptual ability of cognitively impaired patients with the FRP technique.

Thinker invariance: enabling deep neural networks for BCI across more people

OBJECTIVE

Most deep neural networks (DNNs) used as brain computer interfaces (BCI) classifiers are rarely viable for more than one person and are relatively shallow compared to the state-of-the-art in the wider machine learning literature. The goal of this work is to frame these as a unified challenge and reconsider how transfer learning is used to overcome these difficulties.

APPROACH

We present two variations of a holistic approach to transfer learning with DNNs for BCI that rely on a deeper network called TIDNet. Our approaches use multiple subjects for training in the interest of creating a more universal classifier that is applicable for new (unseen) subjects. The first approach is purely subject-invariant and the second targets specific subjects, without loss of generality. We use five publicly accessible datasets covering a range of tasks and compare our approaches to state-of-the-art alternatives in detail.

MAIN RESULTS

We observe that TIDNet in conjunction with our training augmentations is more consistent when compared to shallower baselines, and in some cases exhibits large and significant improvements, for instance motor imagery classification improvements of over 8%. Furthermore, we show that our suggested multi-domain learning (MDL) strategy strongly outperforms simply fine-tuned general models when targeting specific subjects, while remaining more generalizable to still unseen subjects.

SIGNIFICANCE

TIDNet in combination with a data alignment-based training augmentation proves to be a consistent classification approach of single raw trials and can be trained even with the inclusion of corrupted trials. Our MDL strategy calls into question the intuition to fine-tune trained classifiers to new subjects, as it proves simpler and more accurate while remaining general. Furthermore, we show evidence that augmented TIDNet training makes better use of additional subjects, showing continued and greater performance improvement over shallower alternatives, indicating promise for a new subject-invariant paradigm rather than a subject-specific one.