Study of Diffusion Weighted Imaging in presence of noise

Postgraduate Thesis uoadl:1665185 721 Read counter

Unit:
Κατεύθυνση Ιατρική Φυσική-Ακτινοφυσική
Library of the School of Health Sciences
Deposit date:
2017-06-12
Year:
2017
Author:
Birmpas Dimitrios
Supervisors info:
Αστρακάς Λουκάς, Επίκουρος καθηγητής, Ιατρική Φυσική και Κλινική Ακτινολογία, Πανεπιστήμιο Ιωαννίνων
Καλέφ-Εζρά Τζων, Καθηγητής, Ιατρική Φυσική και Κλινική Ακτινολογία, Πανεπιστήμιο Ιωαννίνων
Κουρκουμέλης Νικόλαος, Επίκουρος καθηγητής, Ιατρική Φυσική και Κλινική Ακτινολογία, Πανεπιστήμιο Ιωαννίνων
Original Title:
Εφαρμογή θορύβου στην Μαγνητική Τομογραφία Διάχυσης
Languages:
English
Greek
Translated title:
Study of Diffusion Weighted Imaging in presence of noise
Summary:
Diffusion Weighted Imaging (DWI) is an imaging method, which exploits the random molecular motion of the water in the biological issues due to the diffusing phenomenon. All imaging methods, including DWI, are affected from the noise, originating from multiple sources in the environment of the MRI scanner. As a result of this influence, there is a divergence in the accuracy of the diffusion imaging measurements as well as in the calculation diffusion indices.
The purpose of this master thesis was the study of diffusion weighted imaging in presence of possible real noise sources as well as the theoretical simulation of the diffusing tensor (D) in the presence of multiple levels of white noise in order to assess the noise dependence of diffusion quantification theoretically, but also in practice.
The MRI scanner used was a Philips Intera 1.5 T of the University Hospital of Ioannina along with a water solution phantom, in order to make this study happen.
Processing of Monte Carlo simulation data was done following the development of the scripts algorithms for the automated analysis and the processing of the data in the MATLAB environment.
It was found that there was a statistically important disturbance of the diffusion SNR in the case of open doors of the MRI room (p-value= 0.039), when MRI incompatible were used (p-value< 0.001), in the presence of a bra with metallic additives (p-value= 0.024) and when mobile phones were used outside of the MRI room (p-value<0.001).
The behavior of diffusing indices, which were studied by Pierpaoli and Basser (C. Pierpaoli, P. Basser 1996) was confirmed by Monte Carlo simulations. The diffusing indices used in daily clinical practice were also studied. Moreover, it was shown that the increase of the level of white noise resulted to ellipsoid torsion. The use of multiple gradient directions and powerful gradient fields restricted the torsion effect. In addition, difference between the use of two-dimensional (2D) (D. Jones, P. Basser 2004) and three-dimensional (3D) diffusion tensor was found, supporting the use of 3D modeling for more efficient and more accurate study of the diffusing fractional anisotropy index. Finally it was confirmed that the optimum clinical range of the b-values is between 1.000-2.000 s/mm^2 .
Main subject category:
Health Sciences
Other subject categories:
Algorithms and Theory of Computation
Analysis
Analytical, diagnostic, therapeutic techniques and equipment
Clinical medicine
Particle physics
Keywords:
Diffusion Weighted Imaging, DWI, Monte Carlo, MATLAB, Noise, Diffusion, Eigenvalues, Eigenvectors, Isotropic diffusion, Anisotropic diffusion, white noise
Index:
No
Number of index pages:
0
Contains images:
Yes
Number of references:
43
Number of pages:
76
Dimitrios Birmpas master.pdf.pdf (3 MB) Open in new window