Table 21 Synthetic vs. real data diversity analysis.
Diversity metric | Synthetic data (mean ± SD) | Real data (mean ± SD) | Mean difference (95% CI) | Effect size (Cohen’s d) | t-statistic | p-value | Interpretation |
|---|---|---|---|---|---|---|---|
Structural similarity (SSIM) mean | 0.347 ± 0.156 | 0.298 ± 0.189 | 0.049 (0.031–0.067) | 0.28 | 4.23 | < 0.001 | Higher synthetic diversity |
Perceptual distance (LPIPS) mean | 0.523 ± 0.198 | 0.587 ± 0.214 | -0.064 (-0.089–0.039) | -0.31 | -5.87 | < 0.001 | Slightly lower synthetic diversity |
Feature space entropy | 8.74 ± 1.23 | 9.12 ± 1.45 | -0.38 (-0.62–0.14) | -0.28 | -2.34 | 0.023 | Comparable diversity |
Morphological variation index | 0.612 ± 0.187 | 0.634 ± 0.201 | -0.022 (-0.051-0.007) | -0.11 | -1.42 | 0.156 | No significant difference |
Pathological feature diversity | 0.789 ± 0.134 | 0.823 ± 0.156 | -0.034 (-0.062–0.006) | -0.23 | -1.78 | 0.078 | No significant difference |
Anatomical context variation | 0.845 ± 0.098 | 0.867 ± 0.112 | -0.022 (-0.041–0.003) | -0.21 | -1.19 | 0.234 | No significant difference |
