Digital Signal Processing Reference
In-Depth Information
9.4 Prospects for Pade-optimized MRS for ovarian can-
cer diagnostics
The high resolution of the FPT also could be of benefit for in vivo MRS
investigations, for which, as discussed, poor SNR has been an important ob
stacle hampering progress in ovarian cancer diagnostics via MRS. It has been
suggested that in vivo MRS could become the method of choice for accurate
detection of early stage ovarian cancer, insofar as the current obstacles ham
pering the acquisition of high quality time signals and the subsequent reliable
analysis of spectra as well as their interpretation can be overcome [342]. The
presented results clearly indicate that application of the FPT could be an im
portant step towards achieving the stated goal, particularly insofar as in vitro
and in vivo clinical correlations and histopathology were initially performed
for verification. The metabolite concentration maps could be especially useful
when Padeoptimized MRS becomes implemented in practice. These maps
could then be a handson tool, facilitating not only a deeper understanding
of this novel approach to signal processing, but also rapid and reliable inter
pretation of diagnostically relevant information in the clinical setting.
In the performed analyses, we used noisefree FIDs to set up the fully
controlled standard for the FPT in our initial application of this method to
data within the realm of ovarian cancer diagnostics by MRS. This is method
ologically justified [10]. The next steps will be to extend our analysis to both
noisecorrupted synthesized ovarian data (still wellcontrolled) and to encoded
FIDs such as those from Ref. [343]. These studies are currently underway.
The present results illustrate the capacity of the FPT to resolve and pre
cisely quantify all the physical resonances as encountered in benign versus
cancerous ovarian cystic fluids. Padebased quantification as a parametric es
timation is achieved with an exceedingly small number of signal points. This
has practical implications, because the time signals' exponential tail with
small amplitudes is avoided. Since the tail is embedded in the background
noise, long signal lengths cause severe problems in quantification. Such a
feature of the FPT is particularly advantageous relative to the FFT, which
needs long signals (through which noise is inevitably invoked) in an attempt
to improve resolution 2π/T (T = Nτ).
We have now shown that the fast Pade transform reliably and unequivocally
yields the metabolite concentrations of major importance for distinguishing
benign from malignant ovarian lesions. The FPT achieves this without any
fitting or numerical integration of peak areas. These capabilities of the FPT
are likely to be of benefit for ovarian cancer diagnostics via MRS. Further
pursuit of this avenue of investigation should be a clinically urgent priority.
The urgency is underscored by the fact that early ovarian cancer detection is
a goal which is still elusive and achievement of which would confer a major
survival benefit.
Search WWH ::
Custom Search