Master thesis presentation by Moa Peters
Welcome to attend the master thesis presentation by Moa Peters, MPBME, on February 5th. The title of Moa Peters´work is “Development of an analysis pipeline for magnetoencephalography measurement data using the MNE software package”. Examiner: Justin Schneiderman
Magnetoencephalography (MEG) is a non-invasive, functional neuroimaging method mapping the activity within the brain by measuring the magnetic fields generated by neuronal currents. One great benefit with MEG is its sub-millisecond temporal resolution, enabling better understanding of the high-frequency mechanisms of the brain as compared to common functional neuroimaging methods such as functional magnetic resonance imaging (fMRI) with a temporal resolution around 1 s. There are many challenges in MEG data analysis e.g. the data is often affected by environmental noise and physiological artifacts caused by eye blinks, heartbeats and muscle activity. It is important to put effort and thought into the MEG data analysis pipeline to be able to present the measurement data in a reliable, neurophysiologically correct way for medical experts to interpret and a processing pipeline needs to be tailored for each MEG study based on variables like the hypothesis under test, the measurement equipment used and the experimental protocol.
During this thesis work, a pipeline for MEG data analysis has been developed with focus on preprocessing of measurement data and analysis on sensor level (studying the magnetic fields rather than the computed neuronal activity). The pipeline was developed using the MNE software package and optimized for analyzing event related potentials from somatosensory stimuli. An explorative and iterative method was used, based on analyzing four MEG data sets from pilot measurements within an ongoing medical research project in combination with feedback from medical and technical experts for evaluating and choosing processing methods within the pipeline. Analysis methods discussed in this thesis are e.g. visual inspection of measurement data, filtering and averaging for noise reduction, artifact removal methods including signal space separation, signal space projection and independent component analysis.