Digital Signal Processing Reference
In-Depth Information
active tumor growth was total creatine.
8.1.3.2
Tumor grading based upon a combination of metabolites
via MRS & MRSI
Pattern recognition based upon a larger number of metabolites has also been
used for tumor grading, since it is appreciated that relying upon single metabo
lites can be unreliable [113]. A study of forty eight patients using a 3T scanner
by Di Costanzo et al. [257] corroborates that several metabolites and their ra
tios yielded better accuracy in classifying high versus low grade gliomas. Her
minghaus et al. [249] employed a combination of metabolites to aid in grading
histopathologically confirmed glial brain tumors among ninety patients. With
singlevoxel proton MRS, linear discriminant analysis based upon normalized
NAA, total creatine, choline, lactate and lipid to total Cr ratio was used
to classify tumor grades. Accurate tumor grading was achieved in 77% to
94% of cases. The most serious grading errors occurred in two patients with
Grade I/II who were considered to be Grade III/IV and for three patients
with Grade III/IV who were judged as I/II. These authors [249] consider that
spectral pattern analysis yielded a high level of accuracy with the possibility
of identifying cases that cannot be reliably graded, and “thereby avoids ther
apeutic decisions based on ambiguous data” (p. 79). An automated pattern
recognition system based upon a data base of over 300 tumor spectra and
using several metabolites has been developed within the European Union, to
aid in brain tumor diagnostics with MRS and MRSI. However, tumor grading
using that system cannot yet be considered reliable since there is considerable
overlap due to tumor heterogeneity [113].
8.1.4 Characterization of brain tumors by MRS & MRSI
Brain tumor types can sometimes be identified by using MRS and MRSI [258].
8.1.4.1
Identification of meningiomas
Distinguishing meningiomas, which are usually slow growing from more ag
gressive brain tumors, is most often achieved by MRI and CT. When mor
phologic characteristics are indeterminate, MRS and MRSI have been helpful.
The appearance of alanine at 1.48 ppm on MRS or MRSI is considered by
some authors to be highly suggestive of meningioma [113, 211, 259]. Alanine
was found to be present in fifteen of nineteen patients with meningiomas, but
was not detected in eleven Schwannomas, or in any of the eight metastatic
tumors examined in Ref. [144]. Higher alanine and glutamine - glutamate
best discriminated between twentytwo patients with meningiomas and nine
patients with primitive neuroectodermal tumors in another study [260]. In
another paper by Majos et al. [261], alanine was found to be elevated in atyp
ical meningiomas, compared to other tumors types. In a third paper, these
authors [262] described alanine as the most characteristic resonance of menin
Search WWH ::
Custom Search