direkt zum Inhalt springen

direkt zum Hauptnavigationsmenü

Sie sind hier

TU Berlin

Page Content

Machine Learning

PhD Theses

New tools for electrophysiological data analysis and their application to a working memory study
Citation key Proepper2015
Author Pröpper, Robert
Year 2015
School Technische Universität Berlin
Abstract In order to understand how signals in the brain are related to observed behavior, it is essential to observe the activity of single neurons. Electrophysiology is one of the oldest and most important means to achieve recordings of the voltage fluctuations produced by active neurons. In extracellular electrophysiology, recording electrodes are placed in the intercellular medium and often record from multiple neurons at the same time. A central issue is to determine which neuron produced each recorded spike, solutions for this problem are called spike sorting. The steady improvement in electrophysiological recording techniques requires advances in data analysis: the increasing amounts of recorded data need to be managed and appropriate analysis algorithms for large data sets need to be developed. As experiments grow more complex, interdisciplinary cooperation becomes more important. This necessitates data sharing and collaborative development of analyses. The first part of this thesis is concerned with a solution to the challenges arising from this trend: Spyke Viewer is a software platform for electrophysiological data management and analysis that supports many different data formats in a unified way. It is focused on usability and flexibility, so it is useful to both experimenters who can use it to browse and visualize data and theoreticians who develop new analysis algorithms. With newer electrophysiology techniques, where multiple recording channels capture activity from an increasing amount of neurons, recorded spikes often overlap in time. The resulting waveforms are a particular challenge for spike sorting methods. In this thesis, a spike sorting algorithm that addresses the overlap problem is improved and evaluated on simulated and empirical data. In addition, a complete spike sorting pipeline from raw data to sorted spikes is described. All methods were developed and tested using an empirical data set recorded from the prefrontal cortex of macaque monkeys. The monkeys performed a visual working memory task. Using Spyke Viewer and the improved spike sorting algorithm, large scale analyses on the coding of visual stimuli and experimental conditions were carried out. Reactions of individual neurons were examined and population codes were explored using a variety of decoding methods. Using time-resolved analyses, it was found that the neural coding of all experimental conditions changes quickly over the course of a trial, but sample stimuli and test stimuli elicit very similar neural response patterns at different times during the trial.
Link to publication Download Bibtex entry

Zusatzinformationen / Extras

Quick Access:

Schnellnavigation zur Seite über Nummerneingabe

Auxiliary Functions