Multiplet resonances in MRS data from normal and cancerous prostate - Signal Processing in MRS with Biomedical Applications - page 352

computed concentration was completely correct for the polyamine resonance
at 3.100416 ppm, although Re(ν − k ), Im(ν − k ) and|d − k |were correct to 5, 5 and

4 decimal places, respectively at N P = 500.
For the region between 3.14 ppm to 3.65 ppm, which is spectrally more
dense, none of the reconstructed spectral parameters of the identified reso
nances were fully exact at N P = 500, although the computed concentrations
of scylloinositol (k = 16) and peaks k = 23−25 that were the components of
one of the myoinositol triplets were correct. At N P = 500 the reconstructed
chemical shift, Im(ν − k ) and computed concentration of peak k = 26, the my
oinositol singlet at 4.070235 ppm were completely correct, but not the last
digit of the |d − k

|. The Re(ν − k ) and Im(ν − k ) of the lactate peak k = 27 were
reconstructed correctly at N P = 500 and the computed concentration was
also correct. However, the|d − k |were exact only to 4 decimal places.

Most of the computed concentrations had the correct integer values, except
for the two peaks immediately adjacent to the missed resonances, glycerophos
phocholine (k = 12) and taurine (k = 14); these were both overestimated.
Rather than the correct concentration of 3.13 M/g ww, the concentration
of GPC was computed to be 6.08 M/g ww, at N P = 500 and the recon
structed|d − k

|was 1.112238 au, while the correct value was 0.572736 au. The
concentration of the taurine peak k = 14 was computed as 4.18 M/g ww at
N P = 500, whereas 3.87 M/g ww is the correct value.
Full convergence was attained at N P = 600 for all the reconstructed pa
rameters and computed concentrations for all 27 resonances (bottom panel
(ii) of Table 11.5 ) . At higher partial signal length and the full signal length N
all the reconstructed spectral parameters and computed concentrations were
stable.
Figure 11.4 shows the Padereconstructed absorption component shape spec
tra at N P = 54 and N P = 800 for the normal stromal prostate data. A total
of 16 the 27 peaks were missing at N P = 54 (upper panel (i)). As was seen
for normal glandular prostate, at N P = 54 for the normal stromal tissue,
only at the two extremes of the spectrum were the resonances resolved with
nearly correct heights (lactate and alanine at 1.33 ppm and 1.49 ppm and
myoinositol and lactate at 4.07 ppm and 4.12 ppm, respectively).
In the denser spectral region from≈2.5 ppm to 3.65 ppm, only 7 of the 23
peaks were resolved and the absorption and dispersive modes were mixed. A
singlet near 2.5 ppm was seen rather than the expected citrate doublet. The
citrate doublet near 2.75 ppm appears mainly as a singlet but there is also a
broad component with very small amplitude centered around 2.8 ppm. The
peaks near 3.55 ppm and 3.65 ppm corresponding to myoinositol were seen as
singlets, rather than triplets. There should have been ten peaks between 3.0
ppm and 3.35 ppm, but only two were seen. In the bottom panel (ii) of Fig.
11.4 at N P = 800 all 27 resonances are resolved with the correct peak heights,
including all the multiplets and the overlapping resonances of phosphocholine
at≈3.23 ppm and glycerophosphocholine at≈3.24 ppm. The dispersive

modes are absent at this converged partial signal length.