Fig. 1 | Scientific Reports

Fig. 1

From: Generative adversarial network model to classify human induced pluripotent stem cell-cardiomyocytes based on maturation level

Fig. 1

Overview of the GAN model to improve classification of hiPSC-CM maturation level. Morphology images and contractility recordings (seen domain) are collected to train the GAN model to generate high-quality artificial data composed of cell images and contractile recordings. Relevant synthetic images trained from the seen domain then are mixed with the real data to train the cell classifier to improve its classification accuracy in both the seen and the unseen domains. Fake videos are obtained by training the GAN model with time series images of individual hiPSC-CMs.

Back to article page