How to generate the synthetic spectra

Hi everyone, I met two questions about generating synthetic spectra.

1.SNR is mostly defined as the ratio of peak amplitude to the standard deviation of noise in the frequency domain. How can I add noise in the time domain to achieve a specific value of SNR?

2.Should the summation of metabolite signals also be done in the time domain, similar to the simulation of lipid signals in osprey’s fit_makebasis, where pairwise summation and subsequent correction are performed?

Thanks in advance !

Hi Murph,

For 1), I’d recommend looking into the synthetic data generator in Osprey (specifically lines 376f).

For 2), it doesn’t really matter - the Fourier transform is a linear operator, so the FFT of a sum of functions is equal to the sum of the FFT of the functions.

1 Like