Supplementary Figure 11: Comparison of the results of the algorithm for different sets of three patterns.

Scatter plots showing the correlation of the retrieved amplitudes of the respective component Golgi/giantin (left), F-actin/phalloidin (middle), and DNA/EdU (right). The reference amplitude was obtained using the mean patterns for all three components as shown in Supplementary Fig. 9. The legends indicate the patterns that were altered by plus/minus standard deviation (Supplementary Fig. 9) together with the equation of the linear model describing the data. Correlation coefficients of each data-set with the linear model were larger than 0.99. The dotted lines indicate the shot noise limits A ± square root (A).

The top row (a–c) shows the effect if instead of the reference pattern the pattern plus or minus the standard deviation is used. The effect is strongest for the amplitude of the DNA stain. The linear regression shows a systematic decrease of the amplitude by 6% and 7%, respectively. For the other components, actin and giantin, the effect is in the order of up to 3%. It is usually found, that the pattern showing an average lifetime between the values of the other two components is most sensitive to deviations of the pattern. The middle row (d–f) shows how the amplitude of the respective component responds to changes of the other patterns. The deviations are all in the order of 1% to 3% and even if the two other patterns (bottom row, g–i) are altered, the values do not increase. In conclusion, we see the most pronounced effect, if the pattern of the component deviates from its optimal shape. However, if the deviations are within the standard deviation of the reference pattern, the observed effects are on average in the order of 5%. Furthermore, the analysis shows that the method is non-problematic and shows linear behavior within the scope of this analysis.