Optimized derivative fast Fourier transform: Splitting singlet-appearing resonances to genuine multiplets in ovarian NMR spectra from encoded time signals

Abstract

We address the demanding J-spectroscopy part of nuclear magnetic resonance (NMR) for encoded time signals. In the fast Fourier transform (FFT), the J-coupled multiplets are mostly unresolved even with strong magnetic fields (e.g. 600 MHz, 14.1T). The problem is further exacerbated by minuscule chemical shift bands hosting such multiplets. Derivative estimations might be tried as an alternative strategy. However, too tightly overlapped resonances require higher-order derivative estimations. These, in turn, uncontrollably enhance the reconstruction instabilities. Hence, a robust optimizing stabilizer is needed. It is provided by the optimized derivative fast Fourier transform, which simultaneously increases resolution and reduces noise. We presently demonstrate that higher-orders (up to 15) of this processor can accurately resolve the J-coupled multiplets into their genuine components hidden within the singlet-appearing resonances in the FFT spectra. This is exemplified with the challenging two triplets (taurine, myo-inositol lying within only 0.02 ppm) for time signals encoded by ovarian NMR spectroscopy from a patient’s excised cancerous cyst fluid specimen.