Fig. 4: Comparison of real and synthetic images.
From: Improving 3D deep learning segmentation with biophysically motivated cell synthesis
A Image slices of real and synthetic 3D nuclei images. SimOptiGAN uses a random process for nuclei arrangement, while SimOptiGAN+, Mem2NucGAN-P, and Mem2NucGAN-U incorporate biophysical modeling for a realistic arrangement. B Image slices of real and synthetic membrane signals. The synthetic membrane signal is generated, as described in the “Methods” section, based on the same simulated cell borders used in the nuclei synthesis methods SimOptiGAN+, Mem2NucGAN-P, and Mem2NucGAN-U. Consequently, the membrane signal exhibits a consistent cell arrangement, as demonstrated by the overlay of synthetic membrane with nuclei generated using SimOptiGAN+. C A preview of naively generated data used as the worst example for KID evaluation in (D). These naive images are preliminary outputs of SimOptiGAN, generated by the simulation pipeline before undergoing deep learning optimization. D Comparison of synthetic nuclei images with real counterparts based on the Kernel Inception Distance (KID). Lower scores represent a greater similarity between real and synthetic signals. Scale bar: 50 μm.
