Post-processing of individual signals for de-noising.

In this work, we examine the fluctuation of the intensity and the phase of an NMR signal during repetition of experiments and investigate possibilities of using these information to judge suspicious peaks, whose true colors may be noises or genuine signals. We firstly analyze the intensity and the phase of an NMR signal separately, and show that for the accumulated spectral profile the contribution of the intensity is less than that of the phase. Secondly we show that we can de-noise a noisy spectrum by using the standard deviation of phase at each spectral point. We then compare the de-noising effect of the present approach and that of the phase-covariance method proposed recently, which is an alternative method of appreciating phase distribution. Finally, effects of the dispersion component are discussed.