Multiplet resonances in MRS data from normal and cancerous prostate - Signal Processing in MRS with Biomedical Applications - page 370

limitations, it is still clear that there is a far richer source of spectral infor
mation with which to identify prostate cancer than is currently used with
Fourierbased in vivo MRS.
Padeoptimized MRSI with its capability to unequivocally resolve and quan
tify multiplet resonances and otherwise exceedingly challenging spectra with
many overlapping resonances could undoubtedly provide valuable information
for improving prostate cancer diagnostics.
11.4 Prospects for Pade-optimized MRSI within prostate
cancer diagnostics
As in the other two problemareas within cancer diagnostics, we have used
noisefree FIDs, in order to set up the fullycontrolled standard for the FPT.
The methodological rationale for this strategy has been elaborated [10]. We
are now applying this analysis to both noisecorrupted synthesized data for
the prostate and to encoded FIDs similar to those from Ref. [443] as well as
in vivo MRS and MRSI data from the prostate. These results will be reported
soon.
It should be emphasized that since the time signals from MRSI are precisely
of the same nature as those from MRS, the FPT can also be applied with equal
success to MRSI. We have performed initial applications of the FPT to MRSI
and demonstrated improvements in the resolution of MRSI and mitigation of
Gibbs phenomena [290].
We conclude that the FPT is optimally suited to resolve and quantify the
numerous overlapping resonances, including multiplets of metabolites in this
very di cult area of signal processing in MRS within the realm of prostate
cancer diagnostics.
Only short time signals were needed for this achievement, and, as discussed,
this is a major advantage because free induction decay data become heavily
corrupted with noise at the long total acquisition times as required by the
FFT, which lacks interpolation and extrapolation features.
Herein, we have once again demonstrated that the FPT can unequivocally
disentangle physical from spurious content of the studied time signals.
The many multiplets characteristic of the spectra of healthy and cancerous
prostate represent a major challenge for signal processing. Pade optimization
has been shown here to meet this challenge.
This line of investigation will continue with encoded data from normal,
hypertrophic and cancerous tissue, in vitro and in vivo. We expect that Pade
optimized MRSI will improve the diagnostic accuracy of MRbased modalities.
This could certainly contribute to a more timely and accurate diagnosis of
prostate. Solutions to therapeutic dilemmas might also be forthcoming.