Sign in Subscribe
Machine learning

Breakthrough in Personalized Medicine: Exactech Inc.'s Computed Tomography Image-Based Tool for Segmentation and Quantification of Shoulder Muscles

Strategic Insights

— 1 min read

Researchers at Exactech Inc. have made a significant breakthrough in personalized medicine by developing a computed tomography (CT)-based tool for automated segmentation of deltoid muscles, enabling quantification of radiomic features and muscle fatty infiltration. The tool, which utilizes a machine learning (ML)-based segmentation algorithm, has undergone rigorous validation by three expert shoulder surgeons for 32 unique patients. The validation has demonstrated clinically acceptable performance, with 97% of ML-generated deltoid masks being clinically acceptable.

Key Takeaways:

  • The research aims to conduct shoulder expert validation of a novel deltoid ML auto-segmentation and quantification tool.
  • The study involves training a SwinUnetR-based ML model on labeled CT scans and validating it against 32 unique patients.
  • The validation evaluates the quality of the auto-segmented deltoid images, with each of the three surgeons reviewing the auto-segmented masks relative to CT images.
  • The research found that the ML-generated deltoid masks had a median error in volume and fatty infiltration measurements of 1.1%.
  • The tool has the potential to reliably quantify deltoid muscle size, shape, and quality, facilitating more personalized treatment decision making and evidence-based clinical decisions.
  • The study demonstrates the non-inferiority of the ML model compared to surgeon-corrected deltoid masks and inter-surgeon variation in metrics.
  • The researchers concluded that the CT image analysis tool has clinically acceptable performance for deltoid auto-segmentation.

Statistics:

  • 97% of ML-generated deltoid masks were clinically acceptable.
  • 2 out of 32 ML-generated deltoid masks required major corrections.
  • 1 out of 32 ML-generated deltoid masks was deemed clinically unacceptable.
  • Median error in volume and fatty infiltration measurements using ML-generated masks was 1.1%.
  • The tool has undergone validation by three expert shoulder surgeons for 32 unique patients.

Sources:

  • Clinical Validation of a Computed Tomography Image-Based Machine Learning Model for Segmentation and Quantification of Shoulder Muscles. Algorithms, 2025,18(7):432.
  • Exactech Inc., Gainesville, Florida, United States.
  • Journal of Algorithms, MDPI AG.
  • Hamidreza Rajabzadeh-Oghaz, Exactech Inc., 2320 NW 66th Ct., Gainesville, FL 32653, United States.

Read more