Mitochondrial segmentation and function prediction in live-cell images with deep learning.
Journal:
Nature communications
PMID:
39820041
Abstract
Mitochondrial morphology and function are intrinsically linked, indicating the opportunity to predict functions by analyzing morphological features in live-cell imaging. Herein, we introduce MoDL, a deep learning algorithm for mitochondrial image segmentation and function prediction. Trained on a dataset of 20,000 manually labeled mitochondria from super-resolution (SR) images, MoDL achieves superior segmentation accuracy, enabling comprehensive morphological analysis. Furthermore, MoDL predicts mitochondrial functions by employing an ensemble learning strategy, powered by an extended training dataset of over 100,000 SR images, each annotated with functional data from biochemical assays. By leveraging this large dataset alongside data fine-tuning and retraining, MoDL demonstrates the ability to precisely predict functions of heterogeneous mitochondria from unseen cell types through small sample size training. Our results highlight the MoDL's potential to significantly impact mitochondrial research and drug discovery, illustrating its utility in exploring the complex relationship between mitochondrial form and function within a wide range of biological contexts.