Maximum entropy spectral analysis of 31P NMR signals from human cells

Abstract

Two autoregressive methods, namely Burg and Yule-Walker algorithms, have been applied to 31P NMR free induction decays of packed human tumor cells where the signalto-noise ratio was limited by short duration records. The comparison between FFT and MEM power spectral density points out some advantages of the latter analysis technique. In particular, very good resolution and sensitivity can be reached even with very short data records. The choice of the autoregression order is discussed.