Florin Popescu

Single trial classification of motor imagination using 6 dry EEG electrodes.

Background Brain computer interfaces (BCI) based on electro-encephalography (EEG) have been shown to detect mental states accurately and non-invasively, but the equipment required so far is cumbersome and the resulting signal is difficult to analyze. BCI requires accurate classification of small amplitude brain signal components in single trials from recordings which can be compromised by currents induced by muscle activity. Methodology/Principal Findings A novel EEG cap based on dry electrodes was developed which does not need time-consuming gel application and uses far fewer electrodes than on a standard EEG cap set-up. After optimizing the placement of the 6 dry electrodes through off-line analysis of standard cap experiments, dry cap performance was tested in the context of a well established BCI cursor control paradigm in 5 healthy subjects using analysis methods which do not necessitate user training. The resulting information transfer rate was on average about 30% slower than the standard cap. The potential contribution of involuntary muscle activity artifact to the BCI control signal was found to be inconsequential, while the detected signal was consistent with brain activity originating near the motor cortex. Conclusions/Significance Our study shows that a surprisingly simple and convenient method of brain activity imaging is possible, and that simple and robust analysis techniques exist which discriminate among mental states in single trials. Within 15 minutes the dry BCI device is set-up, calibrated and ready to use. Peak performance matched reported EEG BCI state of the art in one subject. The results promise a practical non-invasive BCI solution for severely paralyzed patients, without the bottleneck of setup effort and limited recording duration that hampers current EEG recording technique. The presented recording method itself, BCI not considered, could significantly widen the use of EEG for emerging applications requiring long-term brain activity and mental state monitoring.

2009 Special Issue: Subject-independent mental state classification in single trials

Current state-of-the-art in Brain Computer Interfacing (BCI) involves tuning classifiers to subject-specific training data acquired from calibration sessions prior to functional BCI use. Using a large database of EEG recordings from 45 subjects, who took part in movement imagination task experiments, we construct an ensemble of classifiers derived from subject-specific temporal and spatial filters. The ensemble is then sparsified using quadratic regression with l(1) regularization such that the final classifier generalizes reliably to data of subjects not included in the ensemble. Our offline results indicate that BCI-naïve users could start real-time BCI use without any prior calibration at only very limited loss of performance.

The Berlin brain-computer interface

The Berlin Brain-Computer Interface (BBCI) uses a machine learning approach to extract subject-specific patterns from high-dimensional EEG-features optimized for revealing the user’s mental state. Classical BCI application are brain actuated tools for patients such as prostheses (see Section 4.1) or mental text entry systems ([2] and see [3,4,5,6] for an overview on BCI). In these applications the BBCI uses natural motor competences of the users and specifically tailored pattern recognition algorithms for detecting the user’s intent. But beyond rehabilitation, there is a wide range of possible applications in which BCI technology is used to monitor other mental states, often even covert ones (see also [7] in the fMRI realm). While this field is still largely unexplored, two examples from our studies are exemplified in Section 4.3 and 4.4.

Single-subject classification of schizophrenia by event-related potentials during selective attention

BACKGROUND Executive dysfunction has repeatedly been proposed as a robust and promising substrate of analytical approaches in the research of neurocognitive markers of schizophrenia. Here, we present a mixed model- and data-driven classification approach by applying a task that targets executive dysfunction in schizophrenia and by investigating relevant event-related potential (ERP) features with machine learning classifiers. METHODS Forty schizophrenic patients and forty matched healthy controls completed the Attention Network Test while an electroencephalogram was recorded. Target-locked N1 and P3 ERP components were constructed and submitted to different classification analyses without a priori hypotheses. Standardized source localization was applied to estimate neural sources of N1 and P3 deficits in schizophrenia. RESULTS We obtained a classification accuracy of 79% using only very few ERP components. Central P3 components following compatible and incompatible trials and right parietal N1 latencies averaged across targets and were sufficient for classification. P3 deficits were associated with anterior cingulate cortex dysfunction, while right posterior current density deficits were observed in schizophrenia during the N1 time frame. CONCLUSIONS The data exemplarily show how automated inference may be applied to classify a pathological state in single subjects without prior knowledge of their diagnoses. While classification accuracy may be optimized by application of other executive paradigms, this approach illustrates the potential of machine learning algorithms for the identification of biomarkers that are independent of clinical assessments. Conversely, data suggest a pathophysiological mechanism that includes early visual and late executive deficits during response inhibition in schizophrenia.

Critical Evaluation of Auditory Event-Related Potential Deficits in Schizophrenia: Evidence From Large-Scale Single-Subject Pattern Classification

Event-related potential (ERP) deficits associated with auditory oddball and click-conditioning paradigms are among the most consistent findings in schizophrenia and are discussed as potential biomarkers. However, it is unclear to what extend these ERP deficits distinguish between schizophrenia patients and healthy controls on a single-subject level, which is of high importance for potential translation to clinical routine. Here, we investigated 144 schizophrenia patients and 144 matched controls with an auditory click-conditioning/oddball paradigm. P50 and N1 gating ratios as well as target-locked N1 and P3 components were submitted to conventional general linear models and to explorative machine learning algorithms. Repeated-measures ANOVAs revealed significant between-group differences for the oddball-locked N1 and P3 components but not for any gating measure. Machine learning-assisted analysis achieved 77.7% balanced classification accuracy using a combination of target-locked N1 and P3 amplitudes as classifiers. The superiority of machine learning over repeated-measures analysis for classifying schizophrenia patients was in the range of about 10% as quantified by receiver operating characteristics. For the first time, our study provides large-scale single-subject classification data on auditory click-conditioning and oddball paradigms in schizophrenia. Although our study exemplifies how automated inference may substantially improve classification accuracy, our data also show that the investigated ERP measures show comparably poor discriminatory properties in single subjects, thus illustrating the need to establish either new analytical approaches for these paradigms or other paradigms to investigate the disorder.

Neurocognitive Pattern Analysis Reveals Classificatory Hierarchy of Attention Deficits in Schizophrenia

Attention deficits, among other cognitive deficits, are frequently observed in schizophrenia. Although valid and reliable neurocognitive tasks have been established to assess attention deficits in schizophrenia, the hierarchical value of those tests as diagnostic discriminants on a single-subject level remains unclear. Thus, much research is devoted to attention deficits that are unlikely to be translated into clinical practice. On the other hand, a clear hierarchy of attention deficits in schizophrenia could considerably aid diagnostic decisions and may prove beneficial for longitudinal monitoring of therapeutic advances. To propose a diagnostic hierarchy of attention deficits in schizophrenia, we investigated several facets of attention in 86 schizophrenia patients and 86 healthy controls using a set of established attention tests. We applied state-of-the-art machine learning algorithms to determine attentive test variables that enable an automated differentiation between schizophrenia patients and healthy controls. After feature preranking, hypothesis building, and hypothesis validation, the polynomial support vector machine classifier achieved a classification accuracy of 90.70% ± 2.9% using psychomotor speed and 3 different attention parameters derived from sustained and divided attention tasks. Our study proposes, to the best of our knowledge, the first hierarchy of attention deficits in schizophrenia by identifying the most discriminative attention parameters among a variety of attention deficits found in schizophrenia patients. Our results offer a starting point for hierarchy building of schizophrenia-associated attention deficits and contribute to translating these concepts into diagnostic and therapeutic practice on a single-subject level.

Performance of novel dry electrode EEG cap for evoked potential and band-power activity detection.

Common applications of electro-encephalography (EEG), in both research and clinical practice, are detection of fast, stimulus locked evoked potential (EP) responses and quantification of slower changes in EEG band-power termed event related synchronization/ desynchronization (ERD/S). These correspond to categorized electrophysiological phenomena associated with various mental states through prior observations. In this paper we present the technical performance of a dry electrode EEG cap previously introduced for demanding ERD/S applications such as brain-computer interfacing, for trial-averaging applications. Results are shown for N100 auditory evoked potential (AEP) as well as ERD/S EEG spectral estimation. The novel cap can allow for EEG experiments which can be set-up within approx. 5 minutes and provide data of similar quality to those traditionally obtained with gel-based caps, without the need for washing subject’s hair or other preparative/follow-up procedures to the experiment.

Detecting Mental States by Machine Learning Techniques: The Berlin Brain-Computer Interface

The Berlin Brain-Computer Interface (BBCI) uses a machine learning approach to extract user-specific patterns from high-dimensional EEG-features optimized for revealing the user’s mental state. Classical BCI applications are brain actuated tools for patients such as prostheses (see Section 4.1) or mental text entry systems ([1] and see [2–5] for an overview on BCI). In these applications, the BBCI uses natural motor skills of the users and specifically tailored pattern recognition algorithms for detecting the user’s intent. But beyond rehabilitation, there is a wide range of possible applications in which BCI technology is used to monitor other mental states, often even covert ones (see also [6] in the fMRI realm). While this field is still largely unexplored, two examples from our studies are exemplified in Sections 4.3 and 4.4.

Algorithmic Design Tool for Integrating Renewable Energy Infrastructures in Buildings: Object Oriented Design for Energy Efficiency

We present a tool which empowers ‘green’ design freedom for architects by presenting ever expanding choices in components and materials and automatizing their configuration and placement. Several time- and resource- consuming initial design iterations are eliminated by optimizing the energetic efficiency of the building in the original draft phase. The smart, efficient, energy producing building of the future can thereby offer increased cost and energy efficiency, security and comfort, without any compromise in style and form - on the contrary, the proposed tool stands to open up a novel palette of creative ‘green’ architectural design elements, which would effectively be co-designed by architects. The proposed algorithmic CAD design tool allows direct integration of renewable sources in the architectural design phase, taking into account local meteorological and solar radiation conditions. Furthermore locally optimized evolution and modification of renewable components integrated into the building’s structure is possible, leveraging an increasingly wide range of possibilities in form, finish and renewable energy generation.

Wagging the Dog: Human vs. Machine Inference of Causality in Visual Sequences

Causal inference among pairs of moving objects in a visual scene is compared between human observers and state-of-the-art methods in Machine Learning for causal inference. It is shown that while humans may perform intuitive and/or reasoned statistical decisions with the same overall level of accuracy as machines, they clearly exhibit biases (or priors) in their judgment and are thus able to make decisions based on much less information than is otherwise required by statistical decision algorithms. While there is no simple explanation for how humans perform this task, connectionist learning structures which implement simple time-delayed correlations (both automatic and deliberative) relying on short-term memory mechanisms may suffice to build complex bottom-up models of the physical world and the interaction therewith.