Image Processing Reference
In-Depth Information
6
Noise Filtering Methods
in MRI
L. Landini, M. Lombardi, and A. Benassi
CONTENTS
6.1 Introduction..............................................................................................169
6.2 The MR Image Model.............................................................................171
6.3 Wavelet-Based Filtering ..........................................................................172
6.4 Adaptive Template Filtering....................................................................174
6.5 Anisotropic Diffusion Filtering ...............................................................177
6.6 Application of Anisotropic Diffusion Filtering ......................................179
References .........................................................................................................183
6.1
INTRODUCTION
It is well known that magnetic resonance imaging (MRI) methods are increasing
diagnostic efficacy due to the recent progress in real-time image acquisition
technology. Such advanced imaging techniques provide access to important
anatomical and functional information through high-speed acquisition and high
spatial resolution.
In MRI applications, there is an intrinsic trade-off between signal-to-noise ratio
(SNR), contrast-to-noise ratio (CNR), and resolution. Depending on specific diag-
nostic tasks, high spatial resolution and high contrast may be required, whereas for
image processing applications, a high SNR is usually necessary because most of
the algorithms are very sensitive to noise. In fact, magnetic resonance (MR) images
often require application of noise filtering techniques before visual inspection or
application of noise-sensitive postprocessing methods such as segmentation algo-
rithms [1]. As a rule, such filtering is desired to significantly decrease image noise
and, simultaneously, to preserve fine image details.
Many efforts were devoted to SNR improvement, including time and spatial
averaging during acquisition. Time averaging has the major advantage that the
SNR increases while the spatial resolution is preserved, provided the imaging
process is stationary; the disadvantage is the time required to perform the exam-
ination. Achieving a high SNR at an elevated spatial resolution may necessitate
169
Search WWH ::




Custom Search