Abstract

Video-based assessment of Parkinson’s disease (PD) motor symptoms could support scalable and more standardized motor evaluation, but progress is limited by the lack of public, task-structured video datasets with clinical annotations. We introduce VAMP-PD (Video Assessment of Motor Performance in Parkinson’s Disease), a single-camera benchmark of 58 subjects, including 43 PD patients and 15 healthy controls, performing six MDS-UPDRS Part~III motor tasks: finger tapping, hand movements, pronation–supination, toe tapping, leg agility, and gait. Videos were assigned consensus MDS-UPDRS severity labels by trained raters under specialist supervision. The dataset includes raw videos, task-level labels, metadata, privacy-preserving processing, and benchmark code to be released under a non-commercial research license. We establish an initial reproducible baseline using MMPose-based keypoint extraction, clinically motivated kinematic features, and a Random Forest classifier with feature selection. Baseline results show that pose-derived features provide signal beyond a majority-class reference, while also highlighting challenges from class imbalance, rare moderate-to-severe examples, and fine-motor pose uncertainty. VAMP-PD is intended to support reproducible benchmarking and future development of clinically grounded video biomarkers for PD motor assessment.

Links to Paper and Supplementary Materials

Main Paper (Open Access Version): https://papers.miccai.org/miccai-2026-sat/paper/AMAI_026.pdf

SharedIt Link: Not yet available

SpringerLink (DOI): Not yet available

Supplementary Material: Not Submitted

Link to Open Review

Open Review Page: https://openreview.net/profile?id=~Mohammadhadi_Saadatnia1

BibTex

@InProceedings{SaaMoh_VAMPPD_MICCAISAT2026,
        author = { Saadatnia, Mohammadhadi AND Haghighi, Ali AND Poureh, Amiratabak AND Baradaranfard, Mahdi AND Saadatnia, Saba AND Ahmadi, Mohammad Hossein AND Tajmirriahi, Amirhossein AND Nasrabadi, Erfan Nasri AND Saadatnia, Mohammad AND Safavi, Saeid},
        title = { { VAMP-PD: A Task-Structured Single-Camera Video Benchmark for MDS-UPDRS Motor Severity Assessment in Parkinson’s Disease } },
        booktitle = {Medical Image Computing and Computer Assisted Intervention -- MICCAI 2026 Workshops and Challenges},
        year = {2026},
        publisher = {Springer Nature Switzerland},
        volume = {LNCS 17273},
        month = {pending},
        page = {pending}
}


back to top