SarcAsM: AI-Powered Sarcomere Analysis

Graphical overview of SarcAsM’s capabilities. Top: Multi-level structural analysis of sarcomere architecture. Bottom: High-precision tracking and analysis of individual and average sarcomere motion.

SarcAsM (Sarcomere Analysis Multitool) is an AI-powered Python package for cardiomyocyte sarcomere analysis. It enables precise multilevel structural and functional assessment of sarcomeres in microscopy images and movies, making it suitable for drug screening, disease phenotyping, and biomechanical studies. Details see our 📄 preprint.

Key Features

SarcAsM comes packed with features designed for comprehensive sarcomere analysis:

• AI-Driven Detection: Robustly identifies sarcomere features in microscopy images using advanced AI.

• In-Depth Structural Analysis: Provides a multi-level examination of sarcomere architecture, including:

  • Z-band morphometrics and lateral alignment.

  • Sarcomere lengths and orientations (sarcomere ‘vectors’).

  • Myofibril characteristics (length, shape).

  • Cell-level myofibril domain organization.

• Precision Motion Tracking: Tracks individual and average sarcomere motion with exceptional accuracy (~20 nm).

• Dynamic Functional Insights: Facilitates detailed analysis of sarcomere contraction and relaxation dynamics.

• Intuitive Standalone Application: Offers an easy-to-use app with an interactive Graphical User Interface (GUI), no coding required.

• Versatile Python API: Includes a comprehensive Python API for custom script development and integration into existing analysis pipelines.

• Efficient Batch Processing: Streamlines high-throughput studies with powerful batch processing capabilities.

Using SarcAsM

SarcAsM offers two convenient ways to analyze your data. Choose the one that best fits your workflow:

• Python Package: For maximum flexibility, performance, and integration into custom analysis pipelines, we recommend the Python package:

  ⚙️ Python Package

  • Ideal for programming-affine users, handling large datasets, and achieving faster processing speeds, especially when leveraging GPU acceleration (if available).

  • Provides a high-level API for scripting and advanced analysis.

• Standalone Application: For a user-friendly experience with a graphical interface, download our ready-to-use application:

  💾 Download App (Windows/MacOS)

  • Note: The standalone application is great for getting started quickly. However, it may take some time to start up initially. The Windows version currently processes data using the CPU and does not support CUDA for GPU acceleration.

Publication and Citation

For a detailed description of SarcAsM, its validation, and example applications, please refer to our preprint. If you use SarcAsM in your research, we kindly ask you to cite this publication:

Haertter, D., Hauke, L., Driehorst, T., Nishi, K., Long, J., Tiburcy, M., Berecic, B., et al. (2025). SarcAsM: AI-Based Multiscale Analysis of Sarcomere Organization and Contractility in Cardiomyocytes. bioRxiv. https://doi.org/10.1101/2025.04.29.650605

Getting Started:

• Installation
  • Installation via pip
  • Installation via GitHub
• Quickstart
  • Sarcomere structure analysis
  • Sarcomere motion analysis
  • Controlling Log Output
• GUI Usage
  • Starting the App
  • Using the App

Developer Guide:

• Development Guide
  • Setting Up Development Environment
  • Running Tests
  • Code Quality
  • Building Documentation
  • Publishing to PyPI
  • Building Standalone Applications
  • Logging
  • Project Structure
  • Version Control
  • Contributing
  • Contact

Tutorials:

• Structural analysis
  • Initialization of SarcAsM object
  • Prediction of sarcomere Z-bands and cell area by deep learning
  • Analysis of cell mask
  • Analysis of sarcomere Z-bands
  • Analysis of sarcomere vectors
  • Analysis of myofibrils
  • Clustering of sarcomere domains
• Motion analysis
  • Automated detection of lines of interest (LOIs) for motion tracking
  • Initialization of Motion object for LOI
  • Full analysis of LOI with default parameters
  • Step-by-step guide
• Batch processing of large data sets
  • Structural analysis
  • Motion analysis
• Data export
  • Structural data of single TIF file
  • Motion data of single LOI
  • Data of multiple TIF-files or LOIs
• Data visualization
  • Example: visualization of structure analysis
  • Example: boxplots of structure features
  • Example: visualization of LOI motion analysis
• Training data generation
  • 1. Creation of training data set and annotation of Z-bands
  • Training of U-Net to predict Z-bands
  • 3. Generation of preliminary training targets
  • 4. Manual curation of M-band masks
  • 5. Generation of final ground truth
• Training of sarcomere U-Net++
  • Prepare and process training data
  • Set training parameters and train
• Training of Z-band 3D U-Net
  • Creation of training data set
  • Training
• Training of ContractionNet
  • Creating training data
  • Training

Sarcomere features:

• Structural Features
• Motion Features

API reference:

• API Reference
  • sarcasm
  • contraction_net

Contact

For questions, requests and issues, please contact us or open an issue on GitHub. For bugs, please append a comprehensive error report.

License

This software is distributed under a custom license and is patent pending (DE 10 2024 112 939.5).

Usage Terms:

• Academic and Non-Commercial Use: The software is free for use in academic research, educational settings, and other non-commercial activities.

• Commercial Use: Commercial use, redistribution, or inclusion in commercial products/services is strictly prohibited without obtaining a separate commercial license.

Commercial Licensing:

For inquiries regarding commercial licensing, please contact:

MBM ScienceBridge GmbH,

Hans-Adolf-Krebs-Weg 1,

37077 Göttingen,

Germany

https://sciencebridge.de/en/

Full License Text:

The complete terms and conditions are available in the LICENSE file included with this software distribution.

Indices and tables

• Index

• Module Index