Breast cancer and non-malignant breast data: Quantification by FPT - Signal Processing in MRS with Biomedical Applications - page 327

resonances. For the malignant data the difference is the most marked between
the total shape spectrum in which the PCho+PE peak at 3.22 ppm is nearly
as large as most abundant resonance, taurine, and the component spectrum
in which phosphoethanolamine and phosphocholine are well delineated and
each much smaller than taurine.
Figure 10.8 displays the converged metabolite maps reconstructed by the
FPT (−) for the normal breast tissue (top panel (i)), fibroadenoma (middle
panel (ii)) and for the breast cancer (bottom panel (iii)). For the normal
breast tissue it is observed that lactate (k = 1), has the largest concentration
(0.5040 M/g ww), slightly higher than β−glucose (k = 7) (0.4500 M/g
ww). The median lactate concentration in the normal breast is about 0.34
that in the fibroadenoma. For the malignant breast, the lactate concentration
is over five times higher than in the fibroadenoma, and is clearly the largest
resonance, nearly three time higher than taurine (k = 8).
In the present analysis the capacity of the FPT to resolve and precisely
quantify very closely overlapping resonances with certainty is clearly demon
strated for the spectrally dense region between 3.21 ppm and 3.23 ppm. This
spectral region encompasses the constituents of total choline: choline at 3.21
ppm, phosphocholine at 3.22 ppm and glycerophosphocholine at 3.23 ppm.
Phosphocholine and phosphoethanolamine are nearly completely overlapping
at 3.22 ppm, separated by only 0.000203 ppm which, as mentioned is about
four times less than the line widths. Nevertheless, at convergence the FPT (−)
exactly reconstructs the input parameters for these two resonances with full
fidelity. As presented earlier with MRS time signals that closely match FIDs
encoded via proton MRS from the brain of a healthy volunteer, the FPT at
convergence also exactly reconstructed all the resonances (25 in that case)
including two that were nearly degenerate. It is also seen herein once again
that convergence of the total shape spectrum does not necessarily imply that
the component spectrum has done likewise.
As discussed, identification and quantification of these constituents of total
choline within the tight spectral region of 3.21 ppm to 3.23 ppm have clinical
relevance for breast cancer diagnostics. In our previous analysis, the ratio of
PCho/GPC was significantly higher in the malignant versus the normal sam
ples [25] of our input data based on Ref. [395]. This is concordant with human
breast cell line research, indicating that malignant transformation is associ
ated with the GPC to PCho switch [307]. Thus, for breast cancer diagnostics,
the proton MRvisible compounds choline (3.21 ppm), phosphocholine (3.22
ppm) and glycerophosphocholine (3.23 ppm) can and should be quantified.
We achieve this by the FPT, rather than summing these three metabolites
via total choline, as is currently done with in vivo MRS.
Precise quantification of the metabolites within this tight spectral region
may also help distinguish fibroadenoma from breast cancer. From the present
input data that rely upon the work of Gribbestad et al. [395], the phospho
choline concentration was approximately 5.7 times lower in the fibroadenoma
than the median phosphocholine concentration for breast cancer.

This was