MSc. Margarita Julia Magadán Méndez in her doctoral thesis states that the use of ICA with cardiac images can be used, showing faster and more precise results than traditional methods for positron emission tomography (PET) studies.

Signals arising in the nature are highly structured. The statistical properties of natural signals are the concern of digital signal processing. The dynamic changes of signals can be seen as statistical properties of the natural signals, the manipulation of these properties in this thesis is used to separate, enhance and analyze specific phenomena of target interest.

Magadán Méndez considers that the use of basic mathematical equations to model the relationship among natural signals is of striking importance. Using these relationships and the Independent component Analysis (ICA) to identify the patterns that characterize the signals of interest is possible, accurate and faster for the visualization of human cardiac tissues than the traditional techniques in positron emission tomography (PET) perfusion studies. The method showed to be feasible for the quantification of the myocardial blood flow.

Margarita Julia Magadán Méndez's dissertation in the field of signal processing, "Signal Separation from Dynamic Data with Independent Component Analysis", will be publicly examined at the Tampere University of Technology (TUT), Faculty of Computing and Electrical Engineering, on Friday 10.10.2008, at 12:00 in the Tietotalo building (Korkeakoulunkatu 1) room TB111.The opponents will be Professor Aapo Hyvärinen (University of Helsinki) and Professor José Fonseca (Universidade Nova de Lisboa, Portugal). Professor Ulla Ruotsalainen from TUT's Institute of Signal Processing will act as a Custos.

Margarita Julia Magadán Méndez is a Mexican native and works as Machine Learning Researcher at Finsor Oy in Espoo.

Further information:

Margarita Julia Magadán Méndez, margarita.magadan@finsor.com