Pose-based tremor type and level analysis for Parkinson’s disease from video - Takeaways - MDSpire

Pose-based tremor type and level analysis for Parkinson’s disease from video

  • By

  • Haozheng Zhang

  • Edmond S. L. Ho

  • Francis Xiatian Zhang

  • Silvia Del Din

  • Hubert P. H. Shum

  • January 18, 2024

  • 0 min

Share

  • 1

    Parkinson's disease affects approximately 10 million people worldwide and is characterized by the loss of dopaminergic neurons, leading to motor dysfunction.

  • 2

    Current PD diagnosis relies on clinical assessments, with an accuracy of 73-84%, which can be influenced by subjective expert opinions.

  • 3

    The proposed video-based deep learning system classifies Parkinson's tremor and estimates tremor severity, achieving 91.3% accuracy in classification.

  • 4

    The system utilizes Eulerian video magnification to enhance subtle tremors and employs AlphaPose for superior 2D pose extraction from videos.

  • 5

    The SPA-PTA classification network incorporates graph neural networks and a pyramidal channel-squeezing-fusion architecture for effective tremor analysis.

Original Source(s)

Pose-based tremor type and level analysis for Parkinson’s disease from video

  • by Haozheng Zhang, Edmond S. L. Ho, Francis Xiatian Zhang, Silvia Del Din, Hubert P. H. Shum

  • January 18, 2024

Related Content

Risk, reward or repetition? New data on human ventral tegmental area stimulation challenge dominant frameworks

AI Scribes: Efficiency for Whom?

Unresolved safety, privacy, and regulatory concerns may limit benefits of AI scribes for physicians and patients.

Assessment of PTeye™ versus FLUOBEAM® LX for parathyroid adenomas: a pilot case–control study