Unbiased measurements of reconstruction fidelity of sparsely sampled magnetic resonance spectra

作者：Wu Qinglin; Coggins Brian E; Zhou Pei^*

来源：Nature Communications, 2016, 7(1): 12281.

DOI：10.1038/ncomms12281

The application of sparse-sampling techniques to NMR data acquisition would benefit from reliable quality measurements for reconstructed spectra. We introduce a pair of noise-normalized measurements, R-work(noise) and R-free(noise), for differentiating inadequate modelling from overfitting. While R-work(noise) and R-free(noise) can be used jointly for methods that do not enforce exact agreement between the back-calculated time domain and the original sparse data, the cross-validation measure R-free(noise) is applicable to all reconstruction algorithms. We show that the fidelity of reconstruction is sensitive to changes in R-free(noise) and that model overfitting results in elevated R-free(noise) and reduced spectral quality.

• 出版日期2016-7