This video explores a powerful approach to analyzing mitochondrial morphology using unsupervised machine-learning techniques. I leverage Principal Component Analysis (PCA) and K-means clustering to transform complex mitochondrial image data into feature spaces that objectively group mitochondria based on their inherent characteristics rather than relying on subjective labels like "large," "small," or "puncta." This approach eliminates arbitrary classifications by identifying more nuanced patterns and subtle variations in mitochondrial structure and function across cell types and experimental conditions. Furthermore, it provides a robust and data-driven way to explore how mitochondrial morphology varies between cell types, ultimately deepening our understanding of cell-specific mitochondrial roles and behaviors.
Updated Jupyter Notebook Hosted On GitHub: https://github.com/farhanaugustine/3D_Mitochondrial_Analysis_Script
**Part 1 (Link):** https://youtu.be/TowK3DMUcaE
