Related Experiment Video
Updated: Jul 7, 2026
13:32
Videomorphometric Analysis of Hypoxic Pulmonary Vasoconstriction of Intra-pulmonary Arteries Using Murine Precision Cut Lung Slices
Published on: January 14, 2014
Spatiotemporal Deep Video-Phenomapping Decodes Microvascular Rarefaction in Middle-Aged and Elder Renovascular
Lingjie Ju1, Ri Ji2, Hong Meng3
1Department of Sonography, Beijing Hospital, National Center for Gerontology, Institute of Geriatric Medicine, Chinese Academy of Medical Sciences and Peking Union Medical College, Beijing 100005, China.
Research (Washington, D.C.)
|July 6, 2026
Summary
AI-powered video analysis identifies distinct kidney blood flow patterns in atherosclerotic renal artery stenosis (ARAS). This new phenotyping accurately predicts kidney events and guides revascularization decisions, avoiding futile treatments.
Area of Science:
- Cardiovascular Imaging
- Artificial Intelligence in Medicine
- Renal Physiology
Background:
- Anatomical stenosis severity in atherosclerotic renal artery stenosis (ARAS) poorly predicts renal outcomes or revascularization benefits.
- Microvascular competence is crucial for assessing renal function in ARAS patients.
Purpose of the Study:
- To develop and validate an AI framework (Renal-Video-AI) for hemodynamic phenotyping in ARAS using contrast-enhanced ultrasound.
- To assess the predictive value of hemodynamic phenotypes for major adverse renal events (MAREs) and treatment response.
Main Methods:
- A self-supervised deep learning framework (Video Swin Transformer with VideoMAE pretraining) was used to extract spatiotemporal hemodynamic features.
- The framework was applied to multi-center human cohorts and a murine model, with validation using spatial transcriptomics.
- Unsupervised phenomapping identified three distinct hemodynamic phenotypes: Preserved, Delayed, and Rarefied.
Main Results:
- The Rarefied phenotype independently predicted MAREs (HR 4.82) and significantly improved clinical model prediction (C-statistic 0.72 to 0.88).
- A significant interaction showed stenting benefited only the Delayed phenotype (absolute risk reduction 12.2%), with no benefit in Preserved or Rarefied phenotypes.
- Spatial transcriptomics revealed hypoxia and pyroptosis in Rarefied tissue; NLRP3-driven pyroptosis was identified as a therapeutic target in a murine model.
Conclusions:
- AI-driven hemodynamic phenotyping reframes ARAS revascularization decisions based on microvascular competence, not just anatomy.
- The study identifies therapeutic futility (Rarefied phenotype) and a treatable window (Delayed phenotype).
- Targeting NLRP3-driven pyroptosis offers a potential therapeutic strategy for ARAS.
