Table 2 Class-wise FID and IL-NIQE. To have a better understanding of the quality of the generated data for each tissue, we calculated the Fréchet Inception Distance (FID) and IL-NIQE score for synthetic (synth.) And real data separately for each tissue type as described in the materials and methods. The FID scores are reported for each tissue class, indicating the dissimilarity between the generated and real images while IL-NIEQ score serves as a measure of image quality, capturing both structural and perceptual aspects.
From: Generating and evaluating synthetic data in digital pathology through diffusion models
Brain | Kidney | Lung | Pancreas | Uterus | |
|---|---|---|---|---|---|
FID | 58.92 | 45.89 | 39.19 | 64.98 | 33.97 |
IL-NIQE (synth. data) | 56.03 | 57.96 | 61.36 | 59.97 | 63.00 |
IL-NIQE (real data) | 32.73 | 33.04 | 34.46 | 30.84 | 35.23 |