UNIVERSITY OF BUCHAREST
FACULTY OF PHYSICS

Guest
2024-11-22 2:00
 HOME     CONFERENCES     SEARCH            LOGIN     NEW USER     IMAGES   


Conference: Bucharest University Faculty of Physics 2015 Meeting


Section: Biophysics; Medical Physics


Title:
Analysis of susceptibility-weighted images using Support Vector Machine in Parkinson's disease


Authors:
Nicoleta-Andreea PASARE, Gabriela NICULESCU, Radu MUTIHAC


Affiliation:
1) University of Bucharest

2) Polytechnic University of Bucharest


E-mail
pasare.nicoleta@yahoo.com


Keywords:


Abstract:
The purpose of the present research work is to differentiate between the stages of Parkinson's disease (PD) by extracting specific features at individual level. To this end, the study was performed by analyzing brain magnetic resonance (MR) susceptibility-weighted images (SWI) by means of support vector machine (SVM) classification algorithms. Magnetic susceptibility images are relatively a new type of MR images within the field of medical imaging modalities. These images display better contrast comparatively to T1- and T2-weighted MR images. Previously reported studies have shown that using SVM classifiers to analyze magnetic susceptibility images can provide significant information about discriminating at individual level between PD and/or other types of neurological diseases. The main purpose of the study is to further refine the analysis of PD patients in order to delineate the initial from the advanced stages of disease at individual level. The results will be statistically validated using a fundamentally different classification approach like hierarchical fuzzy cluster analysis (FCA) and/or spatial independent component analysis (ICA). The data used in the present work were acquired from a group of 25 patients with PD and from a group of 16 healthy subjects using a MR scanner at 1.5 T. The patient group was subsequently divided in two subgroups: 9 advanced stage patients and 16 initial stage patients. The analysis was structured on 4 cases comparing: (i) the two main groups (patients – healthy subjects), (ii) the advanced stage patients and healthy subjects, (iii) the initial stage PD patients and healthy subjects, and (iv) the initial stage PD patients and advanced stage PD patients. The research will mainly promote means for faster diagnosis of PD and more efficient therapy.


Acknowledgement:
I would like to thank Mrs. Mihaela Onu for the data set, without which the study could have not be done.