Skip to main content
AAPM&R 2025 Main banner
JFK
Final Program


Official Program
Icon Legend
This session is not in your schedule.
This session is in your schedule. Click again to remove it.
Presentation Icons
On Demand
On Demand
In-Person
In-Person
Livestream
Livestream
Invite Only
Invite Only
Ticketed
Ticketed
Coffee Break
Coffee Break
Meal
Meal
Reception
Reception
Rate ePoster

Musculoskeletal and Sports Medicine

Exploration of Ultrasound to Detect Myofascial Pain in the Shoulder

Friday, October 24, 2025
9:15 AM - 10:30 AM MT
Location: Session Room - 250 DEF

    Poster Presenter(s)

  • Nathan Katz, DO

    Chief Resident of Recruitment MetroHealth Rehabilitation Institute; Case Western Reserve University
    MetroHealth System/Case Western Reserve University PM&R Program
    Cleveland Heights, Ohio

    • Co-Author(s)

  • JD

    Joana Dilipkumar, Other

    MASc Candidate
    University of Toronto, Toronto Rehabilitation Institute
    Toronto, Ontario, Canada

  • Richard D. Wilson, MD

    Chair
    MetroHealth Rehabilitation Institute
    Cleveland, Ohio

  • DK

    Dinesh Kumbhare, MD

    Professor
    University of Toronto
    Toronto, Ontario, Canada

Disclosure(s):
Nathan Katz, DO: No financial relationships to disclose
Richard D. Wilson, MD: No financial relationships to disclose
Background and/or Objectives: To develop a machine learning algorithm (MLA) to accurately classify participants with and without chronic shoulder pain (CSP), with and without active and latent myofascial trigger points (MTrPs), based on muscle tissue structure of the deltoid and trapezius captured in ultrasound imaging.

Design: Prospective cohort study.

Setting : Urban academic medical centers.

Participants : 70 participants were recruited without CSP, 14 subjects with CSP and active MTrPs, and 5 subjects with CSP and latent MTrPs.

Interventions: Ultrasound images were obtained of participants' trapezius and deltoid musculature of their affected shoulder(s).

Main Outcome Measures: Accuracy of the machine learning model to classify between B-mode ultrasounds with active MTrPs, latent MTrPs and healthy tissue.

Results: 31 texture features were analyzed from B-mode ultrasound images of the muscles and were used to train the machine learning model. Testing the model using similar data resulted in high accuracy overall and within each subgroup. Overall, the model achieved a 91.2% accuracy across all groups, with 88.8%, 95.2% and 89.6% accuracy in classifying images with active MTrPs, latent MTrPs and healthy tissue, respectively.

Conclusions: These results demonstrate that the model can classify between the three groups with high accuracy, and performed the best for latent MTrPs, which clinically, is the most difficult group to identify. Strong classification performance supports the potential of MLAs that use ultrasound-derived texture features as potential biomarkers for myofascial pathophysiology.