Choice of suitable wavelets for MR image processing

Choice of suitable wavelets for MR image processing

Stať ve sborníku v databázi WoS či Scopus

Magnetic nuclear resonance is used in particular as a diagnostic imaging method. Images of selected parts of organs must be of sufficient quality for doctors to be able to not miss any details and to make reliable diagnoses. The detected images are often of low contrast and resolution. They are mainly subjected to noise, whose level depends, among other things, on the level of the signal being detected, local proton density, voxel size, bandwidth, system design, quality of RF coil, and detection parameters [1], [2], [3]. Noise is in the same frequency band as the image spectral components that carry the details. Suppressing noise without knowing its properties must be a compromise between the desired smoothing and improvement of signal to-noise-ratio (SNR) on the one hand and the loss of details on the other hand. If the filtering method is applied in order to remove noise from MR images, a correct choice of individual filtering parameters is important. In the paper, the selection is described of wavelets suitable for improving the quality of MR images of temporomandibular joint [4] and their evaluation according to three criteria: increased SNR, steepness of the change in signal intensity in the image, and the change in contrast.

Magnetic nuclear resonance is used in particular as a diagnostic imaging method. Images of selected parts of organs must be of sufficient quality for doctors to be able to not miss any details and to make reliable diagnoses. The detected images are often of low contrast and resolution. They are mainly subjected to noise, whose level depends, among other things, on the level of the signal being detected, local proton density, voxel size, bandwidth, system design, quality of RF coil, and detection parameters [1], [2], [3]. Noise is in the same frequency band as the image spectral components that carry the details. Suppressing noise without knowing its properties must be a compromise between the desired smoothing and improvement of signal to-noise-ratio (SNR) on the one hand and the loss of details on the other hand. If the filtering method is applied in order to remove noise from MR images, a correct choice of individual filtering parameters is important. In the paper, the selection is described of wavelets suitable for improving the quality of MR images of temporomandibular joint [4] and their evaluation according to three criteria: increased SNR, steepness of the change in signal intensity in the image, and the change in contrast.

SNR, MR imaging, wavelets,

SNR, MR imaging, wavelets,

BARTUŠEK, K.; GESCHEIDTOVÁ, E.

2010

22.03.2010

Tne Electromagnetic Academy

Xian, Čína

978-1-934142-12-7

PIERS 2010 Xian

1565

1568

4

http://hdl.handle.net/