Fig. 5: Effect of scaling bias on multivariate analysis of an OrbiSIMS mouse testis image.
From: Orbitrap noise structure and method for noise unbiased multivariate analysis

a PCA scores (PCs 1-7, 34, 39, 77, 81) of the mouse testis dataset using various scaling methods (purple=positive, green=negative, the absolute maximum in the scale is clipped for better contrast, the scales vary from +/− of this for each PC). b Assumed variance vs. mean plots for different scaling methods compared to WSoR model. c WSoR loadings of PCs 2–7. Intensity distributions of d C5H4N5−, e C25H47SO10−, f C41H79O12S−, g C10H−, h C12H−, i C4H6N3O−, j C4H8N3O2−, k C18H35O2- and l KCl2−. The WsoR PC 7 scores map (and equivalent for other scaling and ion distributions) has been cropped to a smaller number of pixels to facilitate visualisation.