We developed a latent diffusion-based generative model combined with our ZECO Module.
With sparse labeled data for training, our generative model generates 1,000 synthetic images that capture both the morphology and spatial distribution of neurons, guided by segmentation patterns.
