Digital Signal Processing Reference
In-Depth Information
background contribution.
Dilemmas at both long and short echo times concerning assignment and
quantification, especially of overlapping resonances, have also been reported
[235] when clinical MRS findings from patients with high grade gliomas and
solitary brain metastases were compared. The finding that patients with brain
tumor recurrences had elevated lipids and glutamine - glutamate underscores
the importance of proper assignment, as well as quantification.
8.3
Accurate extraction of clinically-relevant metabolite
concentrations for neuro-diagnostics via MRS
8.3.1
Methodological strategy:
The need for standards in
quantification
As discussed, by relying upon the FFT with postprocessing fitting, there are
many discrepancies in the neurooncologic literature. These discrepancies were
particularly flagrant with regard to tumordefining cutpoints for metabolite
ratios that varied widely among various centers. One of the key obstacles
to greater use of MRS and MRSI in neurooncology has been the lack of a
uniform approach to data analysis. Data compatibility is needed for multi
center comparisons that are indispensable for wider routine application of
MRS in neurooncology and elsewhere [306]. Differences in data processing
methods, rather than real differences, are considered as one of the major
contributors to deviations among various clinical results in MRS [259].
At odds with the need for data compatibility is the requirement within
the FFT for fitting, which, as discussed, can lead both to spurious peaks
(overmodeling or overfitting) and true metabolites being undetected (under
modeling or underfitting) and it also renders interstudy comparisons tenu
ous, at best. This is, of course, particularly unacceptable in the area of brain
tumor diagnostics in which the most delicate of clinical decisions are made,
since these decisions require maximal information of the highest possible re
liability. It is precisely here that the advances in computational methods
are uniquely contextualized - for the promise they hold in helping the fight
against one of the most di
cult cancers known to humankind [114].
8.3.2 High-resolution quantification of brain MR signals in
a clinician-friendly format
As reviewed in detail in the previous chapters in this topic, benchmark stud
ies have now been performed showing that the fast Pade transform can yield
highly accurate metabolite concentrations from MRS signals. We have em
phasized the clinical urgency of this information, particularly for cancer diag
Search WWH ::
Custom Search