Dieter Devlaminck

Multisubject learning for common spatial patterns in motor-imagery BCI

Motor-imagery-based brain-computer interfaces (BCIs) commonly use the common spatial pattern filter (CSP) as preprocessing step before feature extraction and classification. The CSP method is a supervised algorithm and therefore needs subject-specific training data for calibration, which is very time consuming to collect. In order to reduce the amount of calibration data that is needed for a new subject, one can apply multitask (from now on called multisubject) machine learning techniques to the preprocessing phase. Here, the goal of multisubject learning is to learn a spatial filter for a new subject based on its own data and that of other subjects. This paper outlines the details of the multitask CSP algorithm and shows results on two data sets. In certain subjects a clear improvement can be seen, especially when the number of training trials is relatively low.

A comparative study of two different regression methods for radiographs in Polish youngsters estimating chronological age on third molars

AIM The aim of this study was to establish a third molar developmental database to model dental age of Polish youngsters, to investigate the rating level of the scores when dividing a year interval into a quarter of a year and to examine sex differences, left-right and upper-lower jaw asymmetry. MATERIAL AND METHODS A cross-sectional sample of 1048 orthopantomograms of 644 females and 404 males aged between 12 and 26 years was investigated using the scoring system of Gleiser and Hunt modified by Köhler. Reference tables according to age were split in a whole year and in quarters of a year using descriptive statistics. The various developmental stages between males and females were analyzed with a paired t-test and the cusum method. Differences in mineralization between the quadrants were analyzed with a two-factor ANOVA and the Duncan post hoc test. The single quadratic and support vector regression were performed to describe the relationship between score and age. RESULTS Dividing age classes in quarters of a year discriminated better between individuals provided that there is a sufficient sampling size for all age classes. The mineralization tempo occurred significantly at a faster rate in males. The maturational events in the upper arch developed significantly at earlier ages for both genders. Obtained chronological age had nearly the same standard error of estimate when calculated with both regression methods. DISCUSSION AND CONCLUSION Comparing the results of the present study with those of other population groups suggests that there are differences in the ageing process of the wisdom tooth. This is the first database of Polish youngsters (15-24 years) with their respective regression equations to yield age estimations.

Adding Machine Vision to Enhance P300 BCI Speller

BCI systems can use additional intelligence such as machine vision or obstacle avoidance routines to provide a goal-oriented approach. This paper describes a P300-based BCI system to control a small robotic arm. On the robot arms gripper a small camera was mounted continuously capturing images from the environment. The software waits for a character from the P300-based BCI speller to turn either clockwise or counterclockwise. Once turning the image detection software automatically searches for the colored object and delivers it to a preset target location. Results showed that the BCI system performed fast and accurate.