Supervised pattern recognition techniques for detecting motor intention of lower limbs in subjects with cerebral palsy

Cerebral Palsy (CP) is one of the major conditions that prevent subjects suffering from having free control over their limbs, currently the use of electroencephalography (EEG) signals to control rehabilitation devices is a very useful alternative. However, these EEG signals are susceptible to noise and a filtering preprocessing is necessary before the feature extraction and classification. There are very good algorithms detecting motor intensities in the upper limbs such as Least Squares Support Vector Machine (LS-SVM) with spectral density characteristics. However, in the present work we propose to determine the algorithms of extraction of characteristics and classification that allow to detect satisfactorily the motor intensities in lower limbs.