Digital Signal Processing Reference
In-Depth Information
readily acquire a deeper grasp of the method, together with acumen in inter
pretation of patterns typical of malignancy versus benign pathologies.
This approach is illustrated in the next part of this topic, with the spectral
features, metabolite concentrations and assignments from MRS, comparing
cancerous and benign pathologies and normal tissue. The benchmark studies
described earlier in this topic, together with the practical implementation
described in this chapter represent a valid, exact, and long awaited approach
to quantification of MRS signals. With this type of implementation, we foresee
that Padeoptimized MRS is set to very soon become a standard diagnostic
tool for various branches oncology, including neurooncology.
8.3.3 Pade-reconstructed lipids in the MR brain spectrum
Let us now return in more detail to the example of lipids within the frequency
range of 0.9 ppm and 1.7 ppm. Depending upon the presence of necrosis,
lipids may or may not be detected in brain tumors. Lipids have generally
been described as absent in normal brain tissue [207, 211]. In the study by
Tarnawski et al. [225] there was no detected lipid whatsoever in brain MRS
recordings from thirty healthy volunteers, and significantly higher lipid/NAA
was found in the tumor bed of fiftyone patients (3.33±3.79) compared to
0.73±0.85 on the contra lateral, noninvolved lobe.
However, in the study of Smith et al. [234], while the lipid to creatine ratio
was lower among 5 healthy volunteer compared to twenty patients with brain
stem neoplasms, two of the five healthy participants had some lipid detected at
1.3 ppm as opposed to eighteen of the twenty patients with tumors. Since the
appearance of lipids often indicates necrosis or disruption of the myelin sheath,
lipids are commonly detected in nonneoplastic intracerebral processes such
as abscesses, hypoxia and infarction, demyelinization, toxoplasmosis, as well
as for about 1 week postepileptic seizures [211, 214]. Lipids at 1.3 ppm
were found in thirteen of fourteen patients with nonneoplastic lesions in the
study by Smith et al. [234]. Lipid concentration may also be an indicator
of brain tumor grade. As stated by Murphy et al. [251]: “the spectra of
glioblastomas are dominated by huge lipid lactate resonances, and with few
other features” (p. 330). Several investigations show that the presence or
greater concentration of lipids is associated with highergrade intracerebral
tumors [239, 242, 251, 252].
In light of this particular example of lipids in the region between 0.9 ppm
and 1.7 ppm, it becomes clear how vital it is to have trustworthy information
about the number of resonances and all their spectral parameters. Continuing
with this example, we can see that from the total shape spectrum (bottom
resonances within the region between 0.9 ppm and 1.7 ppm. However, with the
FPT, it can be unequivocally determined that in the present example there are
exactly four resonances, as per the input parameters defined earlier. All these
parameters: position, width and amplitude are precisely and unequivocally
Search WWH ::
Custom Search