Magnetic resonance in neuro-oncology: Achievements and challenges - Signal Processing in MRS with Biomedical Applications - page 267

with increasing signal length, as clearly demonstrated herein, and in other
examinations of in vivo MRS signals encoded at high magnetic field strength
as seen in chapter 7. The favorable convergence rate of the FPT is one of the
main reasons for the obtained resolution enhancement relative to the FFT.
8.2.2 Unreliable quantifications by fitting FFT spectra
As elaborated in our systematic review [114] through the year 2004, very few of
the metabolite concentrations or their ratios estimated via the FFT unequiv
ocally distinguished brain tumors from normal tissue, nor were these specific
for malignancy. In MRS and MRSI, with conventional Fourier processing, in
fection, demyelinating disorders and infarction often showed spectral changes
that were identical to those of brain tumors [114]. Considering the more recent
data included in the present chapter, this conclusion still holds.
Thus, for example, choline can be elevated not only in brain tumors, but
also in numerous other cerebral pathologies including subacute and chronic
ischemia, infiltrative processes, encephalitis, demyelinization (including the
acute phase and with the chronic plaques of multiple sclerosis), organizing
hematoma, Alzheimer's disease, Down's syndrome, as well as depression, the
reactive astrocytosis of epilepsy, to name a few [211].
Low NAA can reflect loss of neurons with infiltration of brain tissue by
tumor, but as a marker of neuronal viability and density, NAA can also be
decreased in almost any brain insult [211]. On the other hand, as discussed
earlier in this chapter, choline can be low in very small brain tumors in adults,
in necrotic brain tumors or otherwise containing mixed tissues, as well as in
brain tumors in children [211, 221]. Levels of NAA can also sometimes be
normal in small brain tumors. Even alanine, which has been found to be
helpful for identifying meningioma can be elevated in processes such as brain
abscess and neurocystocercosis, as well [211].
The cutpoints for metabolite ratios and choice of the denominator metabo
lite used to indicate the presence of brain tumors vary substantially among
different investigators. For example, McKnight et al. [215] used a cutoff
value of 2.5 for the choline to NAA ratio, reporting that biopsy samples con
taining tumor (Grade II to Grade IV gliomas) were distinguished from those
with normal, edematous, gliotic or necrotic tissue with 90% sensitivity and
86% specificity, comparing preoperative MRSI and histopathology. These au
thors also reported that up to half of the T 2 hyperintense regions outside the
contrastenhancing lesion contained areas with choline to NAA ratios > 2.5.
However, choline to creatine ratios≥1.7 were reported by Murphy et al. [251]
to be associated with unequivocal tumor presence, and that ratios≤1.3 were
seen in normal brain tissue. Among thirtyone patients with malignant brain
tumors, all regions with confirmed cancer showed significant choline levels and
a choline to NAA ratio > 1.3 in an investigation by Vigneron et al. [216].
In the study of Smith et al. [234], choline to creatine ratios in the brain
of healthy persons were found to be as high as 2.14, while choline to creatine