Denoise

In any experimental technique, the distinction between signals and noise is a real factor, and this is particularly the case with NMR because of its low inherent sensitivity. The application of the so-called ‘spatial filters’ has been proposed for the efficient denoising of images as well as for 1D and 2D NMR data sets.

Mnova includes a number of those filters aimed at reducing the noise level while maintaining the original spectral resolution.

Comparison of the MMWF* and eNL-means filter with a synthetic 2D spectrum 10% Gaussian noise (top-left). Top-right spectrum is the original noiseless spectrum whilst the two spectra at the bottom of the figure are the denoised spectra using MMWF* (bottom-right) and eNL-Means (bottom-left)

Enhanced Non-Local Means

This is an algorithm based on the Non-Local Means denoised procedure which exploits the property that many structural similarities are present in different parts of the spectrum and that are not necessarily located in a local neighborhood. This algorithm is complemented by the application of a hybrid median algorithm used to remove potential spikes in the resulting NL-means denoised algorithm.

The blocksize is the default mode as it gives the highest denoising power. As a drawback, it is much slower than the other two modes (pixelwise and fast pixelwise).

Total Variation: denoising algorithm based on Total Variation by Chambolle's projection. Futher information about this algorithm can be found here.