Related Experiment Video
Updated: Jan 16, 2026
07:33
Non-invasive Skeletal Muscle Quantification in Small Animals Using Micro-computed Tomography
Published on: November 8, 2024
832
A deep learning algorithm for automatic 3D segmentation and quantification of hamstrings musculotendon injury from
Lara Riem1, Olivia DuCharme1, Ashley Coggins1
1Springbok Analytics, 100 West South Street, Suite 1E, Charlottesville, VA, 22902, USA.
Scientific Reports
|September 29, 2025
Summary
Artificial intelligence (AI) models accurately quantify hamstring strain edema using MRI scans. This AI-driven approach correlates with injury severity, aiding in athlete recovery and reducing reinjury rates.
Area of Science:
- Sports Medicine
- Radiology
- Biomedical Engineering
Background:
- Hamstring strain injuries are prevalent in high-velocity sports, leading to significant missed playing time and high reinjury rates.
- Current clinical grading of hamstring injuries, like the British Athletics Muscle Injury Classification (BAMIC), relies on semiqualitative assessment of edema, which can be subjective.
- Accurate evaluation of injury severity and location is crucial for guiding effective athlete rehabilitation and return-to-sport protocols.
Purpose of the Study:
- To develop and validate automated artificial intelligence (AI) models for segmenting edema and hamstring muscle/tendon structures from MRI scans.
- To quantify edema volume and its impact on hamstring muscles using AI-derived measurements.
- To assess the correlation between AI-based quantification of edema and established clinical injury grading systems (BAMIC).
Main Methods:
- AI models were trained to automatically segment edema (T2-weighted MRI) and hamstring muscle/tendon structures (T1-weighted MRI).
- MR scans were acquired from collegiate football athletes at the time of injury and upon return to sport.
- AI models performed volumetric, length, and cross-sectional area (CSA) measurements of segmented structures and subregions, comparing favorably against ground-truth segmentations.
Main Results:
- AI volumetric measurements showed high correlation with ground truth for edema (R=0.97), hamstring muscles (R≥0.99), and hamstring tendon (R≥0.42).
- Increased edema volume and percentage of muscle affected by edema significantly correlated with higher clinical BAMIC injury grades (p<0.05).
- The developed AI models demonstrated robust performance in segmenting and quantifying key injury parameters.
Conclusions:
- AI-based quantification of edema in hamstring strain injuries provides objective and reliable measurements.
- These AI-derived metrics effectively reflect varying levels of injury severity, supporting clinical validity.
- This approach offers a promising tool for enhancing the assessment and management of hamstring injuries in athletes.
