Advanced Statistical Machine Learning Methods for the Analysis of Neurophysiologic Data with Medical Application

Abstract

Transcranial magnetic stimulation procedures use a magnetic field to carry a short-lasting electrical current pulse into the brain, where it stimulates neurons, particularly in superficial regions of the cerebral cortex. It is a powerfull tool to calculate several parameters related to the intracortical excitability and inhibition of the motor cortex. The cortical silent period (CSP), evoked by magnetic stimulation, corresponds to the suppression of muscle activity for a short period after a muscle response to a magnetic stimulation. The duration of the CSP is paramount to assess intracortical inhibition, and it is known to be correlated with the prognosis of stroke patients’ motor ability. Current mechanisms to estimate the duration of the CSP are mostly based on the analysis of raw electromyographical (EMG) signal and they are very sensitive to the presence of noise. This master thesis is devoted to the analysis of the EMG signal of stroke patients under rehabilitation. The use of advanced statistical machine learning techniques that behave robustly in the presence of noise for this analysis allows us to accurately estimate signal parameters such as the CSP. The research reported in this thesis provides us with a first evidence about their applicability in other areas of neuroscience.