Signal Separation with A Priori Knowledge Using Sparse Representation - Advances in Heuristic Signal Processing and Applications - page 323

Fig. 14.7 Separation results of in vivo human brain 1H MRS spectra: ( a ) the observed spectrum
from tumor tissue (zone 1) and the estimated baseline ( dashed line ); ( b ) the observed spectrum
from normal tissue in the contralateral hemisphere (zone 2) and the estimated baseline ( dashed
line ); ( c ) the estimated resonances with the spectrum in ( a ); ( d ) the estimated resonances with the
spectrum in ( b )
and the corresponding estimation results of the two spectra. The peak area ratios of
(choline+creatine+polyamine) to citrate (Cho+Cr+PA/Cit) were calculated as final
quantitation results. For the spectra in Fig. 14.8 (a) and Fig. 14.8 (b), the values of
(Cho+Cr+PA/Cit) are respectively 1.18 and 0.39. As a published study [ 20 ] has
shown that cancerous prostate tissues associate to relatively high (Cho+Cr+PA/Cit)
value, we can conclude that the spectrum in Fig. 14.8 (a) corresponds to cancerous
tissue. The conclusion accords with the final biopsy diagnosis.
14.4 Summary
Single channel signal separation with a priori knowledge using sparse representation
can relax in a certain degree the disjoint condition concerning source signals in
the time-frequency domain. The basis of this kind of methods is decomposing the