PCA based background removal
What is it about?
The paper explores background removal based on the results of a statistical method (PCA) compared with a range of techniques that estimate background on individual signals. PCA summarises all the common signals that reccur in a dataset. Because it is based on the whole dataset, it is possible to define all different sources of backgrounds much more accurately and reproducibly. This leads to considerable improvement in performance of analytical models created from the data.
Why is it important?
The concept of estimating and correcting backgrounds on PCA loadings rather than individual signals is a paradigm shift. Most methods either correct background on individual signals or use statistics to identify and exclude low information components. This paper demonstrates that actively manipulating the PCA loadings is a powerful tool for cleaning up data and achieving high levels of accuracy and reproducibility in analysis.
The following have contributed to this page: Rene Beattie
In partnership with: