Digital Signal Processing Reference
In-Depth Information
ratios as low as 0.36 could be seen at sites of brain tumors. As presented,
more recent investigations have also used varying cutpoints of metabolite
ratios to define the presence of brain tumors. For example, choline to NAA
ratios > 1.9 at long TE were used to define tumorous from pseudotumorous
lesions in a study of eightyfour patients with solid brain masses by Majos et
al. [228]. On the other hand, in a study by Hourani et al. [227] a cutpoint
of NAA to choline of 0.61 (whose reciprocal is 1.64) was used to distinguish
various grades of brain tumor from nonmalignant lesions.
Metabolite ratios can be affected by many confounding factors including
cancer treatment itself. The use of ratios is problematic for accurate assign
ment of spectral changes to specific disease processes [295], since they often
vary for reasons unrelated to the oncologic process of interest. Notably, the
ratio of choline to NAA, upon which detection of brain tumor is heavily based
in proton MRS, may also be due to loss of NAA, which is seen in a wide vari
ety of neurological disorders including epilepsy, multiple sclerosis, as well as in
cerebrovascular accidents. Metabolite ratios can be affected by contamination
from adjacent tissues. High NAA can occur with contamination from adjacent
tissues within brain tumors [216]. Metabolite ratios can also be affected by
regional differences. There is a wide variation in metabolite ratios in normal
brain, especially comparing white and grey matter [211, 259].
We have discussed earlier in this chapter that treatmentrelated changes
can affect metabolite ratios. Alteration in the ratios of choline to NAA
and to creatine may also occur as a result of subclinical neurotoxicity re
lated to chemotherapy, possibly due to neuronal loss and/or inhibition of cell
metabolism, as well as to radiation effects [291, 296]. Moreover, metabolite
ratios are affected by measurement parameters. Since different metabolites
have different relaxation times, metabolite ratios are dependent upon echo
time. Changes in TE were noted to impact upon the usefulness of metabolite
ratios for, e.g., distinguishing various grades of brain tumors [252].
Earlier in this chapter, we reviewed the data showing that brain tumor
grading has been substantially aided by in vivo MRS. Nevertheless, there
are quite a few contradictory findings in the literature. Thus, for example,
while higher mean choline levels and lower mean NAA levels are generally
associated with highgrade tumors, there are large standard deviations, which
are considered to preclude accurate tumor grading [216]. This is due, at
least in part, to the fact that highgrade brain neoplasms frequently contain
necrosis, such that choline levels and their ratios to NAA and creatine may be
similar to or even less than those of lowgrade brain tumors [211, 242, 252, 297].
In one of these studies [297], choline concentrations were reported to be
significantly lower in 13 highgrade compared to 10 lowgrade gliomas. We
also noted that pattern recognition based upon a larger number of metabolites
has also been used for tumor grading, because it was recognized that relying
upon single metabolites can be unreliable [113]. However, tumor grading
using that system was not considered su
ciently reliable either, since there
was substantial overlap of findings due to tumor heterogeneity [113].
Search WWH ::
Custom Search