Prony analysis based parameter estimation of an NMR signal of blood plasma for cancer detection

Researchers have investigated the correlation between statistical differences in the aliphatic region of NMR processed blood plasma and malignant cancer. Previous efforts used nonlinear curve-fitting algorithms to indirectly estimate the nine components of interest in the methyl and methylene peaks of this region. A different method is described in this paper for directly estimating the nine components. A preprocessing algorithm is used to remove all components outside the aliphatic region. Prony´s method and Kumaresan and Tufts´ modification are then run, using an order reduction algorithm to identify the compounds of interest. These methods are applied to an NMR plasma sample and the results are compared to the nonlinear curve-fitting algorithm, showing a close relationship