Finally, these marks/masks in the qBEI image were transferred/overlaid directly to the elemental maps (Fig. 2). A general normalization of the XRF count rates for acquisition time and synchrotron-ring current of 100 mA was performed. The XRF intensities of Pb, Zn, and Sr were further corrected for variations in XRF intensities caused by slight changes in the measurement selleck inhibitor setup between different maps, samples and synchrotron sessions, so that the Pb, Zn, and Sr XRF-intensities between all the maps can be directly compared and treated as measures of elemental content. For this purpose an average factor

K (see formula (1)) was evaluated for each map, expressing the mean ratio between Ca as measured by qBEI (wt.% Ca) and Ca as measured by SR μ-XRF(cpsCa). Thus, the multiplication of the SR μ-XRF cps values of Pb, Zn, and Sr from the individual maps with the corresponding K factors leads to a correction/normalization of all the maps based on the absolute Ca values as obtained by qBEI method. equation(1) K=1n∑i=1nwt.%CaicpsCai Formula 1: K = mean selleckchem normalization factor of one

SR μ-XRF map, wt.%Cai = averaged Ca concentration of mineralized bone matrix ROIi measured by qBEI, cpsCai = mean Ca-Kα fluorescence intensity of mineralized bone matrix ROIi, n = number of the mineralized bone matrix ROIs of the respective map. For each sample the medians of the normalized count rates of Ca, Zn, Pb and Sr for the mineralized Olopatadine bone matrix and

the cement line ROIs were calculated. The levels of significance of the differences between mineralized bone matrix and cement lines were tested with the non-parametric Mann–Whitney test for each sample separately. For this purpose all evaluated mineralized bone matrix and cement line ROIs of the respective sample were used. The number of mineralized bone matrix and cement line ROIs was different for all samples. The number of cement line ROIs was larger for all samples. To evaluate the changes in count rate ratios between cement lines and mineralized bone matrix the Wilcoxon signed rank test with the hypothetical median value 1 (= equal elemental distribution) was used. The significance of the correlation between Ca content and trace element levels of all evaluated mineralized bone matrix ROIs of all samples (n = 402) was tested with the non-parametric Spearman’s test. Differences or correlations with p < 0.05 were considered significant. It has to be emphasized that the spot size of the confocal SR μ-XRF setup is about 5 times wider than the width of the cement lines. Thus the levels of trace elements in the cement lines presented in the following are actually a huge underestimate of the real levels of trace elements (see details in “Limitations” section). In Fig.