Data Processing Methods in MRS - Advance Image Processing in Magnetic Resonance Imaging - page 403

area of the resonance peak is proportional to the concentration of the detected
substance. The quantification of these parameters leads to metabolic tissues'
characterization [1,2].
Since MRS has been applied to humans, H-MRS has attracted much

attention. The reason is that a proton is the most sensitive and stable nucleus,
and hydrogenous atoms are largely diffused in living tissues [3]. Nowadays,
thanks to the recent technical advances in MR instrumentation,

1

H-MRS is
routinely applied in clinical settings, especially in brain study, where it has
been documented to be effective in the diagnosis, prognosis, and treatment
selection of cerebral tumors [4,5], cerebral ischemia [6], epilepsy [7,8], and
multiple sclerosis [9,10]. Changes in the

1

H-MRS resonance patterns were
observed between normal brain and cerebral tumors, with potential applica-
tions for the grading and classification of different tumor types [11,12]. The
metabolism of cerebral ischemia shows both acute and chronic changes, with
relevant implications from a pathophysiological and a therapeutic point of
view [16,17]. Finally, a few studies describe the possibility of investigating
the metabolic characteristics of multiple sclerosis (MS) lesions classified in
acute, subacute, and chronic cases, using MR spectroscopy [9,10,18]. In
addition to brain studies, other body tissues have been investigated using
MRS, including prostate, liver, and muscle. In particular,

1

P-MRS has been
used in the diagnosis of muscular disease such as McArdle's syndrome [19]
and Duchenne dystrophy [20].
Because of its clinical importance, the processing of

31

MRS signals and
the extraction of the relevant information is not a trivial task. Major problems
that may limit the theoretical potentiality of the technique include the narrow
chemical shift range of

in vivo

1

H signals, which requires a precise shimming of the B

0

field, and the presence of unwanted water and lipids contributions, which over-
whelm the small metabolites of medical interest [3,13]. Also, the presence of
severe phase distortions [14] and consistent overlaps among spectral peaks [15]
make it difficult to quantify the parameters of interest, especially in a clinical
environment where short echo time and low-intensity magnetic fields are
employed. Finally, good shimming and correct suppression of water contribution
usually depend on the intervention of the experimenters, thus limiting the repeat-
ability of the study [3].
For these reasons, there is a need for robust and reliable signal processing
methods that make it possible to extract the relevant FID information. The meth-
ods should be fast, automatic, and operator independent.
In this chapter, we briefly introduce the basic principles of the methodology
available for advanced quantitative FID analysis and for the extraction of
metabolite parameters. It is not within the scope of this chapter to provide an
exhaustive description of the wide range of signal processing methods pro-
posed for MRS study. Rather, it presents some introductory concepts of signal
processing that may be fruitfully applied for FID analysis, focusing the reader's
attention on the potentiality and the flexibility of these techniques in metabolite