Baseline Correction
Baseline Correction |
|
|
Spectra suffering from high background intensity may need baseline correction to improve their quality prior to analysis. The UVIR plugin of Mnova offers two algorithms that can handle even very complex baselines: •Asymmetric Least Squares (AsLS) Baseline Correction •Multipoint Baseline Correction
Asymmetric Least Squares Baseline Correction:1,2 Combines a smoother with asymmetric weighting of deviations from the smooth trend to get an effective baseline estimator. There are two algorithm parameters helping to optimize the results; “AsLS Logλ” (from 0.1 to 12; being 6 the default value) and “AsLS asymmetry” (0.000001 - 0.999999; being 0.000001 the default value). 1 P.H. Eilers, H.F. Boelens, Leiden University Medical Centre Report 1, 5, 2005. 2 Journal of The Institute of Electronics and Information Engineers Vol.53, NO.3, 2016.
Checking the 'show Baseline' box will display the baseline in your spectral window with the option to compute the initial values for assymmetry (for PLS algorithms) or smoothing factor (for SNIP algorithm) and ratio parameters. Unchecking the 'Show Baseline' box will apply the baseline correction with the selected parameters.
In the IAsLeast Squares algorithm, the parameters Asymmetry and Ratio have similar influence in the construction of the baseline, though the Ratio tends to have a less critical influence (i.e. in the number of needed internal iterations for convergence). For further information about his algorithm, please check the reference below: https://doi.org/10.1364/AO.404863 For further information about SNIP algorithm, please check this article: Multipoint Baseline Correction: This method provides a way of modeling the baseline by selecting a well-distributed set of points that fall on the baseline and then interpolating between those points to complete the model. Multipoint series baseline correction is a very useful algorithm of spectral (pre)processing that eliminates intensive spectral backgrounds, such as a fluorescence signal in Raman spectra, in a simple and efficient way. Although the algorithm can be applied to single spectra, its main value is the possibility of processing spectral series with essentially different individual baselines using a set common baseline points.
After having clicked on the 'Multipoint Baseline Correction button'; you will get a dialog box which will allow you to put down points along the baseline (by clicking on the 'Pick Point' button) to help the program to find the correct coefficients for the baseline correction equation that it will subtract from the spectrum. If you want to remove any undesired point, just double click on it (click and drag if you want to change the location of the point). Let´s see the functionality of each button of the ''Multipoint Baseline Correction' dialog box: 'Pick Baseline points': Allows the user to pick the points. To remove any selected point, just double click on it. 'Automatic': Automatically add points for the baseline correction. 'Pick Borders': Allows the user to easily pick both the first and the latest points, if they need a correction. 'Clear Points': Click on this button to delete all the selected points (restart your work). 'Preview': To see a preview of the baseline correction prior to apply the changes. 'Apply': To apply the correction. 'Free Selection': Check this box to be able to pick points anywhere (otherwise their vertical positions are strictly defined by the baseline curve). 'RMS Calculation span (points): This option takes spectrum noise into account at the baseline construction. The value of 1 means that any picked point is picked exactly on the spectral curve. Otherwise, the point will be set into a vertical position corresponding to a virtually smoothed spectrum. The virtual smoothing is performed by simple averaging of points within a window where the selected point stays in the centrum and the present option indicates the number of neighbors from each side to be included into the averaging window. Therefore, the full window width is 2*n+1 (n is the option value). For example, if you set this option to 3, the ordinates of each selected baseline point will be calculated as an average of ordinates of the spectral points within a window of 7 points with the selected point in the middle. 'Function': from this drop-down menu, we can select the baseline construction function on the basis of selected points: Linear Segments: Simply connects the points with lines. Cubic Splines: Uses the method of splines to connect the points. Witthaker: Fits the baseline points using the Witthaker Smoother (the points do not generally belong to the resulting baseline!). |
