Asynchronous Brain Computer Interfaces (BCI) al-low system activation at free will. This scenario would be desirable for potential users such as patients with motor disabilities, however, there are still several limitations such as long calibration times and overall performance. Different approaches to improve these systems explore diverse linear electroencephalography (EEG) features in order to describe cognitive states. Alternatively, the present study proposes the use of recurrence quantification analysis (RQA), which provides indices about the dynamical behaviour of non-linear systems. We evaluated the accuracy of continuous detection of mental calculation and idle state using RQA features from EEG recordings. Among the different classification methods that were tested, random forest (RF) classifier resulted in the best performance. From 35 selected RQA features, 12 healthy subjects achieved mean accuracy of 0.924 ±0.048, whereas with 150 selected features mean accuracy was 0.977 ±0.020. These results suggest that RQA features are adequate for continuous task episode detection in asynchronous BCI. However, a restriction in the number of these features, conceived for improving computational costs and time, does impact detection performance.

Electroencephalography (EEG) is the most frequently used technique to monitor functional activity of the brain. It has been widely employed in brain-computer interfaces based on the detection of P300 potentials. However, the P300 waves often contain physiological and non-physiological artifacts such as steady state visually evoked potential, power line or environment noise. The aim of this work is to eliminate undesirable periodic independent components from EEG, in order to enhance the P300 wave. The proposed method combines independent component analysis with a suitable selection of the most representative P300 components according to power features estimated from time measures using Parseval's theorem. The results show statistical differences (p<;0.001) between the power spectral densities of raw and restored EEG, after Parseval-based component elimination. Additionally, the comparison of P300 latencies between raw and filtered EEG, showed statistical differences (p<;0.001). Our findings suggest that this method can be helpful to eliminate undesirable components with significant narrow-band power, in order to preserve information required to enhance the P300 potential.

Brain magnetic resonance imaging segmentation is accomplished in this work by applying nonparametric density estimation, using the mean shift algorithm in the joint spatial-range domain. The quality of the class boundaries is improved by including an edge confidence map, that represents the confidence of truly being in the presence of a border between adjacent regions; an adjacency graph is then constructed with the labeled regions, and analyzed and pruned to merge adjacent regions. In order to assign image regions to a cerebral tissue type, a spatial normalization between image data and standard probability maps is carried out, so that for each structure a maximum a posteriori probability criterion is applied. The method was applied to synthetic and real images, keeping all parameters constant throughout the process for each type of data. The combination of region segmentation and edge detection proved to be a robust technique, as adequate clusters were automatically identified, regardless of the noise level and bias. In a comparison with reference segmentations, average Tanimoto indexes of 0.90-0.99 were obtained for synthetic data and of 0.59-0.99 for real data, considering gray matter, white matter, and background.

An automated approach for template-free identification of partially occluded objects is presented. The contour of each relevant object in the analyzed scene is modeled with an approximating polygon whose edges are then projected into the Hough space. A structurally adaptive self-organizing map neural network generates clusters of collinear and/or parallel edges, which are used as the basis for identifying the partially occluded objects within each polygonal approximation. Results on a number of cases under different conditions are provided.

Describes an automated approach for analyzing complex scenes obtained from digitized scanning electron microscopy (SEM) images of lung bio-dissolution preparations. Such images are used in the evaluation of airborne fiberglass distributions. They present varying densities of isolated, overlapping, and crossing fibers and other objects that should be individually identified and measured. A polygonal approximation model for each detected object in the scene is used to obtain a simplified description. After detecting complex objects, such as crossing fibers, an unsupervised neural network is used to resolve the ambiguities and separate the objects. The method performs satisfactorily under different conditions of fiber density and background clutter.