Breast cancer and non-malignant breast data: Quantification by FPT - Signal Processing in MRS with Biomedical Applications - page 330

percentagewise a greater difference than for phosphoethanolamine or glyc
erophosphocholine. Clearly, from these input data based upon median values
from a fairly small number of breast cancer samples and only one fibroade
noma, no definitive conclusions can be drawn about which metabolites best
detect breast cancer and distinguish it most clearly from normal mammary
tissue or benign lesions. Nevertheless, our multiple logistic regression analysis
[21, 22, 25] of these data from Ref. [395] indicates that only lactate showed
100% diagnostic accuracy both with and without inclusion of the fibroade
noma. Thus, the lactate concentration from the present input data [395] was
lower in the fibroadenoma than in all the individual breast cancer samples.
10.5 Prospects for Pade-optimized MRS for breast can-
cer diagnostics
Our results for Padereconstruction of MRS data employed noisefree FIDs,
since we wanted to set up the fullycontrolled standard for the FPT in the
case of the initial application of this method to data within the realm of
breast cancer diagnostics by MRS. As elaborated, this is methodologically
justified. We are now taking the next steps to extend our analysis to both
noisecorrupted synthesized data (still wellcontrolled) and to encoded FIDs
similar to those from Ref. [395] as well as in vivo MRS data from the breast.
We therefore conclude that the demonstrated advantages of the FPT could
definitely be of benefit for breast cancer diagnostics via MRS. This line of in
vestigation will now continue with encoded data from benign and malignant
breast tissue, in vitro and in vivo. Clinical correlations among the in vitro
and in vivo findings and histopathology will be vital for initial verification
of the studies in breast cancer diagnostics. We foresee that Padeoptimized
MRS will reduce the false positive rates of MRbased modalities and further
improve their sensitivity. Once this is achieved, and since MR entails no
ionizing radiation, new horizons open up for screening and early detection.
This could be especially important for risk groups. Thus, for example, we
envision the possibility that Padeoptimized MRS could be used with greater
surveillance frequency among younger women with deleterious BRCA muta
tions, LiFraumeni syndrome and for other women at increased breast cancer
risk. It should also be recalled that there are promising data concerning
the application of MRS for early assessment of response of breast cancer to
chemotherapy [118, 386, 394]. Some of these were based upon attempts to
estimate total choline concentrations from Fourier total shape spectra. It can
be anticipated that Padebased quantification of the components of choline
and other diagnostically important metabolites might further contribute to
this clinically vital step of optimizing therapeutic strategies for breast cancer.