An instantaneous frequency based algorithm for the processing of NMR data

This paper deals with the processing of Nuclear magnetic resonance spectroscopy signals. An instantaneous frequency based algorithm (the IGN-IF method) was recently proposed and shown to be capable of resolving weak signals that are not discernible using the traditional FFT based techniques. In this work we develop a modified IGN-IF (MIGN-IF) algorithm that is adaptive to the decay rates of the signal components. In particular, we develop expressions that permit the parameters of the Gabor filterbank and estimation step to be determined from the decay factor. Simulation results show significant performance improvement of the MIG-IF over the original IGN-IF algorithm. We also compare the MIGN-IF method to the information theoretic criteria matrix pencil (ITCMP) method. Although the estimation variance of the ITCMP method is slightly lower than that of the MIGN-IF algorithm, the ITCMP has a much heavier computational cost and significantly higher breakdown threshold, making it practical only at high SNRs and for short data records.