12
emoryhealthcare.org/ortho Dr. Daniel Whittingslow, PhD Dr. Xerogeanes, MD Dr. Inan, PhD Joint Sounds as a Biomarker of Knee Injury Status

Joint Sounds as a Biomarker of Knee Injury Status · 2020. 8. 16. · emoryhealthcare.org/ortho Dr. Daniel Whittingslow, PhD. Dr. Xerogeanes, MD. Dr. Inan, PhD. Joint Sounds as a

  • Upload
    others

  • View
    0

  • Download
    0

Embed Size (px)

Citation preview

  • emoryhealthcare.org/ortho

    Dr. Daniel Whittingslow, PhDDr. Xerogeanes, MD

    Dr. Inan, PhD

    Joint Sounds as a Biomarker of Knee Injury

    Status

  • emoryhealthcare.org/ortho

    BACKGROUND

    Fig 1. Proposed model of knee AE generation

    A. B.

    C.D.

    E.

  • emoryhealthcare.org/ortho

    History Physical Exam Laboratory Tests Imaging

    Rehabilitation & Recovery

    vs.Surgical Medical

    BACKGROUND

  • emoryhealthcare.org/ortho

    vs.Surgical Medical

    History Physical Exam Laboratory Tests Imaging

    Rehabilitation & Recovery

    BACKGROUND

  • emoryhealthcare.org/ortho

    Acoustic emissions (AEs) generated during joint articulation contain clinically-relevant information pertaining to the underlying health of joints. Can these sounds differentiate healthy knees from knees with acute meniscal tears?

    Joint sounds

    OBJECTIVE

  • emoryhealthcare.org/ortho

    Cadaver Model:

    Figure 2. Recording Setup

    METHODOLOGY

    Testing Stages: Baseline Sham Tear RemovalN=20

    Clinical Data:

    Recruitment ongoing 11 meniscus tears Contralateral leg as control

  • emoryhealthcare.org/ortho

    Fig 3. Leave-One-Subject-Out Cross Validation (LOSO-CV) is used to train the model

    Predicted ClassProbabilities

    Joint Health Score

    Fig 4. Logistic Function and classification hyperplane

    METHODOLOGY

  • emoryhealthcare.org/ortho

    Model Trained using Cadaver Data

    Model Outputs Joint Health Score for each patient

    METHODOLOGY

  • emoryhealthcare.org/ortho

    CADAVER RESULTSStudy Group: Accuracy Precision (PPV) Recall (Sensitivity) AUCLateral Tear –Cadaver 80% 75% 90% 91.5%

    Medial Tear – Cadaver 90% 83% 100% 97.5%

    Fig 5. Cadaver Lateral Meniscus Tear Comparisons Fig 6. Cadaver Medial Meniscus Tear Comparisons

  • emoryhealthcare.org/ortho

    CLINICAL RESULTSStudy Group: Accuracy Precision (PPV) Recall (Sensitivity) AUCHealthy Leg vs Meniscus Tear Leg 77% 69% 100% 100%

    N=11Fig 7. Human Subjects Meniscus Tear Comparisons Fig 8. Leg Injury Classification Accuracy

  • emoryhealthcare.org/ortho

    Cycle based-AE analysis can accurately classify injury status in a cadaver model

    These patterns appear to be preserved in vivo.

    With further research, the proposed joint health score could be used as a biomarker of joint health.

    CONCLUSION

    Figure 8. Acute Injury Classification Models Diagram

  • emoryhealthcare.org/ortho

    THANK YOU!

    • Daniel Whittingslow– [email protected]

    Slide Number 1BackgroundBackgroundBackgroundSlide Number 5Slide Number 6Slide Number 7Slide Number 8Slide Number 9Slide Number 10ConclusionThank you!