Quantitative Evaluation of Tendon Gliding Sounds and Their Classification Using Deep Learning Models.

Area of Science:

• Biomedical Engineering
• Musculoskeletal Health
• Artificial Intelligence in Medicine

Background:

• Tendon disorders manifest with distinct acoustic and tactile features during movement.
• Artificial intelligence (AI) and machine learning (ML) show promise in medical diagnostics, especially pattern recognition.
• Digital stethoscopes offer a potential method for capturing subtle acoustic signatures of tendon function.

Purpose of the Study:

• To develop and evaluate a deep learning (DL) model for classifying tendon gliding sounds.
• To determine if biomechanical differences in tendons create identifiable acoustic signatures.
• To explore the potential of AI for non-invasive diagnosis of musculoskeletal conditions.

Main Methods:

• Tendon gliding sounds were recorded from the thumb and index finger of healthy volunteers using a digital stethoscope.
• Recordings were transformed into spectrograms for analysis of frequency characteristics.
• Deep learning models were employed to classify sounds based on spectrogram features.

Main Results:

• The DL model achieved high classification accuracy in distinguishing between different tendon sounds.
• Spectrogram analysis revealed distinct acoustic signatures related to tendon biomechanics.
• Findings suggest AI can effectively differentiate tendon sound patterns.

Conclusions:

• AI-based models demonstrate potential as accurate, non-invasive tools for diagnosing tendon disorders.
• This approach may aid in the early detection of conditions like tenosynovitis and carpal tunnel syndrome.
• Further research could validate this method for clinical application in musculoskeletal health.