Metabolite-based wavelets for analyzing magnetic resonance spectroscopic signals

We analyze Magnetic Resonance Spectroscopic signals by the continuous wavelet transform. Instead of the standard (Morlet) wavelet, we introduce a new class of wavelets, derived from the metabolite data themselves, using the properly normalized autocorrelation function of each signal. This allows to detect without ambiguity the presence of a given metabolite in a signal consisting of many different components.