Fig. 2
Comparison of FocAn-, FoCo- and manual foci counting in the same image data set, consisting of a random mixture of non-irradiated and irradiated (2 Gy) DK-MG and SNB19 cells (N = 100 cells). The insets in a-c depict the regions of interest (either midsection a and c, or whole nucleus b) in which foci were counted. The data acquired by FocAn was plotted against data of either a manual point-and-click approach (a and b) or FoCo-based data (c). The dashed lines in a-c illustrate ideal 1:1 relationships between the compared counting methods. The linear regressions to the data (solid lines in a and c) deviate only slightly from the 1:1 relation (for detail, see text). Comparison of the total 2D foci numbers (FN2D, a and c) also reveals little difference (~ 2–3%) between the applied methods (d). The 3D foci number per nucleus (FN3D) determined with FocAn exceeds the number of manually detected foci by ~ 14% (b and d). Moreover, with increasing foci number (i.e. FN3D > ~ 50), FocAn yielded increasingly higher FN3D values as compared to manual counting (b), as illustrated in (b) by the upwardly curved linear-quadratic fit (solid line) diverging from the 1:1 relationship (dashed line). The bars shown in (d) are relative differences in foci numbers with respect to those detected by FocAn, calculated as RelDiff = (FN-FNFocAn)/FNFocAn) × 100%