Generative Adversarial Network Model to Classify Human Induced Pluripotent Stem Cell-Cardiomyocytes based on...

Area of Science:

• Cardiology
• Biotechnology
• Computational Biology

Background:

• Accurate classification of human cardiomyocytes is crucial for understanding cellular structure and function.
• Limited scale and diversity of real experimental image data hinder computational analysis throughput.
• Human induced pluripotent stem cell derived cardiomyocytes (hiPSC-CMs) are a key model for studying cardiac development and disease.

Approach:

• Developed a generative adversarial network (GAN)-based method to generate synthetic image data of hiPSC-CMs.
• Trained the GAN model using optical measurements of hiPSC-CMs cultured on micropatterned hydrogels and control groups.
• Integrated synthetic data with real experimental data to enhance classification models.

Key Points:

• The GAN model successfully replicates true features from real cardiomyocyte image data.
• Inclusion of synthetic data significantly improves cell classification accuracy compared to using real data alone.
• The proposed GAN-based approach outperformed conventional machine learning algorithms in data generalization and classification accuracy.

Conclusions:

• Synthetic data generation using GANs is a valuable tool for overcoming data limitations in biological research.
• This method enhances the classification accuracy and computational analysis of cellular structure and function.
• The study highlights the importance of integrating synthetic data to address challenges of limited sample sizes.