Institutional Repository
Technical University of Crete
Technical University of Crete
EN
|
EL
| URI: | http://purl.tuc.gr/dl/dias/7594D53E-44B3-439A-A256-662FB4BE0872 | ||
| Year | 2024 | ||
| Type of Item | Diploma Work | ||
| License |
|
||
| Bibliographic Citation | Ioannis Karvounakis, "Analysis of functional magnetic resonance imaging data on the Fisher-Shannon information plane", Diploma Work, School of Electrical and Computer Engineering, Technical University of Crete, Chania, Greece, 2024 https://doi.org/10.26233/heallink.tuc.101875 | ||
| Appears in Collections |
The Fisher-Shannon-Information-Plane (FSIP) represents a novel and compact framework for analyzing and distinguishing time series data. FSIP is based on only two parameters: the Fisher-Information-Measure (FIM) and Shannon-Entropy-Power (SEP). FSIP proves particularly useful in differentiating time series and categorizing them into distinct signal types, providing a powerful tool for data exploration. This thesis investigates FSIP's application to data derived from a variety of probability distributions, including the Normal, Power Exponential, Student-t, Gamma, Weibull, Log-Normal, and Uniform models, focusing on how scale and shape parameters influence the FSIP representation. The methodology for FSIP estimation involves estimating the probability density function (PDF) of a time series by means of Kernel-Density-Estimation (KDE), enabling precise computation of FIM and SEP. KDE can be effectively applied to time series with complex correlations. Functional Magnetic Resonance Imaging (fMRI) provides a compelling context for applying FSIP. However, in this thesis FSIP application to fMRI data is treated primarily as a case study. The thesis explores the analysis of Blood Oxygen Level Dependent (BOLD) response for synthetic and real fMRI data, using FSIP to identify patterns and classify signals of brain activity time series. The findings highlight FSIP's versatility and its potential as a reliable method for distinguishing intricate signal patterns in biomedical data and other fields.
Copyright © DIAS 2013
