Manuel Abbafati

Evaluation of the performances of different P300 based brain-computer interfaces by means of the efficiency metric

The aim of this paper is to show how to use the Efficiency, a brain-computer interface (BCI) performance indicator, to evaluate the performances of a wide range of BCI systems. Unlike the most used metrics in the BCI research field, the Efficiency takes into account the penalties and the strategies to recover errors and this makes it a reliable instrument to describe the behavior of real BCIs. The Efficiency is compared with the accuracy and the information transfer rate, both in the Wolpaw and Nykopp definitions. The comparison covers four widely used classifiers and different stimulation sequences. Results show that the Efficiency is able to predict if the communication will not be possible, because the time spent to correct mistakes is longer than the time needed to generate a correct selection, and therefore it provides a much more realistic evaluation of a system. It can also be easily adapted to evaluate different applications, so it reveals a more general and versatile indicator for BCI systems.

Single-epoch analysis of interleaved evoked potentials and fMRI responses during steady-state visual stimulation

OBJECTIVE Aim of the study was to record BOLD-fMRI interleaved with evoked potentials for single-epochs of visual stimulation and to investigate the possible relationship between these two measures. METHODS Sparse recording of fMRI and EEG allowed us to measure BOLD responses and evoked potentials on an epoch-by-epoch basis. To obtain robust estimates of evoked potentials, we used blocks of contrast-reversing visual stimuli eliciting steady-state visual evoked potentials (SSVEPs). For each block we acquired one volume of fMRI data and we then tested for co-variations between SSVEPs and fMRI signals. Our analyses tested for frequency-specific co-variation between the two measurements that could not be explained by the mere presence/absence of the visual stimulation. RESULTS Condition-specific single-epoch SSVEPs and fMRI responses were observed at occipital sites. Combined SSVEPs-fMRI analysis at the single-epoch level did not reveal any significant correlation between the two recordings. However, both signals contained stimulation-specific linear decreases that may relate to neuronal habituation. CONCLUSIONS Our findings demonstrate robust estimation of single-epoch evoked potentials and fMRI responses during interleaved recording, using visual steady-state stimulation. SIGNIFICANCE Single-epochs analysis of evoked potentials and fMRI signals is feasible for interleaved SSVEPs-fMRI recordings.

Introducing NPXLab 2010: A tool for the analysis and optimization of P300 based brain-computer interfaces

Brain-Computer Interfaces (BCI) are emerging as a powerful tool for providing an alternative way of communication and environment control to severely disabled people. Among these systems, P300-based BCIs are widely diffused as they are easy to manage and do not require a training for the subjects. These systems, however, are still too slow so that they are actually used only by those patients that are unable to control any muscle. It is possible to improve their performances, but many different analyses need to be performed. Here a set of tools are described for the analysis and optimization of this class of BCI protocols that allow increasing the performances of such systems.

A UML model for the description of different brain-computer interface systems

BCI research lacks a universal descriptive language among labs and a unique standard model for the description of BCI systems. This results in a serious problem in comparing performances of different BCI processes and in unifying tools and resources. In such a view we implemented a Unified Modeling Language (UML) model for the description virtually of any BCI protocol and we demonstrated that it can be successfully applied to the most common ones such as P300, μ-rhythms, SCP, SSVEP, fMRI. Finally we illustrated the advantages in utilizing a standard terminology for BCIs and how the same basic structure can be successfully adopted for the implementation of new systems.

Moving Towards a Hardware Implementation of the Independent Component Analysis for Brain Computer Interfaces

Brain computer interface (BCI) systems implement a communication path between human users and the external environment by translating physiological signals directly acquired from the brain into commands toward external peripherals. A lot of protocols have been implemented in the BCI field and a lot of analytical techniques and algorithms on the signals have been tested to improve the reliability of the information extracted from signals and then the performances of BCI systems. Independent component analysis (ICA) revealed to be a useful tool for analyzing data as it allows the separation of the signals in some independent sources which carry information about the different components of the signals themselves. However ICA is computationally expensive and some efforts should be done in order to maximize its results in terms of time spent for the analysis. A hardware implementation is now discussed which makes the ICA more useful for the online analysis typical of BCI systems.

Mental task recognition based on SVM classification

In this paper, a brain/computer interface is proposed. The aim of this work is the recognition of the will of a human being, without the need of detecting the movement of any muscle. Disabled people could take, of course, most important advantages from this kind of sensor system, but it could also be useful in many other situations where arms and legs could not be used or a brain-computer interface is required to give commands. In order to achieve the above results, a prerequisite has been that of developing a system capable of recognizing and classifying four kind of tasks: thinking to move the right hand, thinking to move the left hand, performing a simple mathematical operation, and thinking to a carol. The data set exploited in the training and test phase of the system has been acquired by means of 61 electrodes and it is formed by time series subsequently transformed to the frequency domain, in order to obtain the power spectrum. For every electrode we have 128 frequency channels. The classification algorithm that we used is the Support Vector Machine (SVM).

A new visual feed-back modality for the reduction of artifacts in mu-rhythm based brain-computer interfaces

A common problem in EEG recording sessions is that results can be heavily contaminated by artifacts. One of the main reasons is that eyes movements generate a noise signal that superimpose to the data. In some BCI protocols the user has generally to control the movement of a cursor on a PC screen by self-regulating his/her mu-rhythm. In general this requires the user to move the eyes to follow the same cursor, thus intrinsically generating a huge amount of noise. To overcome this problem a new feedback modality has been developed, which is able to dramatically reduce the artifacts as it does not require subjects to move their eyes.

Efficiency of a BCI system in a visual P300 protocol with different stimulation intervals

P300-based BCI systems are widely diffused as they are easy to manage and do not require a training for the subjects. A lot of P300-related BCI researches have been proposed to date and each of them tries to focus on different aspects of these systems. In this paper a study regarding the efficiency of a P300 BCI system, implemented with two different inter-stimulus intervals is presented; the aim is to furnish some guidelines for the implementation of P300 BCI systems and to demonstrate, by means of a comparison among the two conditions, that the efficiency is a reliable indicator of the performances of these systems.

Mental Tasks Recognition for a Brain/Computer Interface

In this paper, a brain/computer interface is proposed. The aim of this work is the recognition of the will of a human being, without the need of detecting the movement of any muscle. Disabled people could take, of course, most important advantages from this kind of sensor system, but it could also be useful in many other situations where arms and legs could not be used or a brain-computer interface is required to give commands. In order to achieve the above results, a prerequisite has been that of developing a system capable of recognizing and classifying four kind of tasks: thinking to move the right hand, thinking to move the left hand, performing a simple mathematical operation, and thinking to a carol.