Detailed statistical comparison of CD62p cluster distributions within two clots.

First level analysis is depicted in (a-c). (a) shows the detailed analysis of the full samples. Density and curvature distributions of all clusters for each cluster dimension were compared at one time using a KS-/WX-Test. The results indicate a similarity of clusters for all cluster dimensions. As an indicator, an aggregated p-value is determined using the weighting t-Norm-AND. The p-value (black dashed line) remains above the critical p-value area (orange bar) proving the similarity hypothesis for all cluster dimensions. The corresponding similarities are sim_M = 0.87/0.82 for density and curvature, respectively. (b) depicts the aggregated p-value calculated with the weighting t-Norm-AND (blue) and confidence interval (black dashed line). A value of sim_M = 0.82 is determined. The full-sample analysis does not give the full parameter range of the analysis, just a brief overview on cluster density and curvature similarity. (d) depicts 100 curves–representing relative density of 100-resamples each including 2000 points (green)–used for the bootstrap based comparison of the cluster/curvature density distribution for both samples. In black, the averaged values for all dimensions are represented. The detailed analysis for the bootstrap data shown in (d) was performed similarly to the analysis in (a). (c) shows the KS-/WX-test compared data regarding cluster density and curvature for all given cluster dimensions. As for the detailed analysis (d), the KS/WX-tests (blue/red) as well as the t-Norm-AND aggregated p-values are plotted (black). Additionally, the boundaries of the confidence interval are displayed (magenta). The comparative analysis of the bootstrap derived data shows a strong similarity between the samples (sim_M = 0.93 for cluster density, sim_M = 0.95 for curvature). (e-i) depict the second level analysis. (e) shows the generalized density/curvature p-values determined via a mean-cross comparison. Herein, p-values of the mean-cross comparison (blue) and lower/upper confidential boundaries (black dashed line) are shown. In general, a significance similarity between the samples is visible (sim_M = 0.92 for mean cluster density and sim_M = 0.94 for curvature distributions). (h) shows the aggregated p-values for cluster curvature and density distributions (blue). The aggregated p-values show high similarities sim_L = 1. (f) depicts the calculated Ripley’s–K/-H function values for the bootstrapped data (green) for all cluster dimensions. Using H-Ripley´, the crucial dimension (representative for the largest difference (clustering-caused) between the sample and the Poisson distribution) can be determined (for the two images, the largest difference occurs for clusters sizes of 220 nm). (g) depicts the comparison of the K-/H-Ripley function values at a generalized level (second level statistics). The p-values of the mean-cross comparison (blue) and lower/upper confidential bounds (black dashed line) are shown. The cross-comparison of the statistical cluster-data analysis also indicates similarity of the samples. (i) shows the result of a pairwise KS-test on the distribution of average cluster density and distribution of average cluster curvature for both samples. We observe only a slight dissimilarity for cluster dimensions between 200 nm– 400 nm and 600 nm– 900 nm. This result confirms the similarity of both samples (sim_M = 0.58).

(TIF)