Quantitative statistical analysis, optimization and noise reduction of atomic resolved electron energy loss spectrum images.

In this work we investigate methods of statistical processing and background fitting of atomic resolution electron energy loss spectrum image (SI) data. Application of principal component analysis to SI data has been analyzed in terms of the spectral signal-to-noise ratio (SNR) and was found to improve both the spectral SNR and its standard deviation over the SI, though only the latter was found to improve significantly and consistently across all data sets analyzed. The influence of the number of principal components used in the reconstructed data set on the SNR and resultant elemental maps has been analyzed and the experimental results are compared to theoretical calculations.