UC Santa Cruz researchers have developed a novel method to address the challenge of limited annotated data for training artificial intelligence (AI) models in single-cell segmentation. Their approach involves using a microscopy image generation AI model to create realistic images of single cells, termed “synthetic data,” which are then utilized to train AI models for improved single-cell segmentation. The software, called cGAN-Seg, generates annotated and labeled images that closely resemble real microscopy images, facilitating the training of segmentation models. This advancement holds promise for accelerating cell behavior studies, disease detection, and drug discovery efforts, as manually segmenting cells from backgrounds is time-consuming and labor-intensive.
Keywords: AI, synthetic cell images, enhanced microscopy analysis, single-cell segmentation
