| abstract |
Using regression analysis, n-dimensional spectral and imaging data from sources including Nuclear Magnetic Resonance or Electron Spin Resonance spectrometers, can be automatically analyzed and the data acquisition equipment or the data itself can be corrected for various determined types of distortion. The method allows correction of, for example, signal drift, sample saturation, removal of phase, baseline and shim distortions and allows removal of unwanted signals from the data. From ratios of determined parameter and error values signals can be distinguished from noise-related responses and using Monte Carlo simulations the corresponding signal detection probabilities can be derived. With the model of an n-dimensional signal, a parametric description of the dataset can be obtained with parameter and error values which truly reflect signal overlap and mutual parameter correlations. From this parametric description higher-level interpretations of the data, such as spin systems and related signal patterns, can be derived efficiently. |