Beyond the SNR-resolution uncertainty principle: Optimized derivative fast Fourier transform for NMR diagnostics in medicine

Abstract The present study is on proton magnetic resonance spectroscopy (MRS), as it applies to tumor diagnostics in cancer precision medicine. The goal with the employed patients’ data, subjected to shape estimations alone with no fitting, is to reconstruct self-contained quantitative information of diagnostic relevance. This can be accomplished by proper evaluation of physical metabolites, especially cancer biomarkers (lactates, cholines, citrates,...). Such information is completely opaque in the encoded time signals, but can be transparent in the frequency domain. The optimized derivative fast Fourier transform (dFFT) can meet the challenge. The thorniest stumbling blocks in MRS are abundant overlapping resonances of low resolution and poor signal-to-noise ratio (SNR). Attempts to increase resolution are marred by decreased SNR. The long-sought strategy of MRS, simultaneous improvement of resolution and SNR, is achievable by the optimized dFFT. With the implied aid to decision-making, this is illustrated for ovarian MRS data encoded from benign and malignant human biofluid samples.