Jove
Visualize
Contact Us
JoVE
x logofacebook logolinkedin logoyoutube logo
ABOUT JoVE
OverviewLeadershipBlogJoVE Help Center
AUTHORS
Publishing ProcessEditorial BoardScope & PoliciesPeer ReviewFAQSubmit
LIBRARIANS
TestimonialsSubscriptionsAccessResourcesLibrary Advisory BoardFAQ
RESEARCH
JoVE JournalMethods CollectionsJoVE Encyclopedia of ExperimentsArchive
EDUCATION
JoVE CoreJoVE BusinessJoVE Science EducationJoVE Lab ManualFaculty Resource CenterFaculty Site
Terms & Conditions of Use
Privacy Policy
Policies

Related Experiment Videos

Uncertainty-Aware Vision-Language Learning Improves Chest X-Ray Retrieval.

Ahmad Elallaf, Yuktha Priya Masupalli, Gongbo Liang

    IEEE Pulse
    |May 15, 2026
    PubMed
    Summary
    This summary is machine-generated.

    Related Concept Videos

    Uncertainty: Overview00:59

    Uncertainty: Overview

    In analytical chemistry, we often perform repetitive measurements to detect and minimize inaccuracies caused by both determinate and indeterminate errors. Despite the cares we take, the presence of random errors means that repeated measurements almost never have exactly the same magnitude. The collective difference between these measurements - observed values - and the estimated or expected value is called uncertainty. Uncertainty is conventionally written after the estimated or expected value.

    You might also read

    Related Articles

    Articles linked to this work by shared authors, journal, and citation graph.

    Sort by
    Same author

    Human induced pluripotent stem cell-derived chimeric antigen receptor-macrophages eradicate IL-13Rα2-positive solid tumors.

    The Journal of pathology·2026
    Same author

    Multi-Scale Self-Supervised Consistency Training for Trustworthy Medical Imaging Classification.

    Annual International Conference of the IEEE Engineering in Medicine and Biology Society. IEEE Engineering in Medicine and Biology Society. Annual International Conference·2025
    Same author

    TMEM35B as a novel biomarker for diagnosing gliomas.

    Biomarkers in medicine·2024
    Same author

    H-Net: Heterogeneous Neural Network for Multi-Classification of Neuropsychiatric Disorders.

    IEEE journal of biomedical and health informatics·2024
    Same author

    Self-Supervised Learning Application on COVID-19 Chest X-ray Image Classification Using Masked AutoEncoder.

    Bioengineering (Basel, Switzerland)·2023
    Same author

    Dynamic Image for 3D MRI Image Alzheimer's Disease Classification.

    Computer vision - ECCV ... : ... European Conference on Computer Vision : proceedings. European Conference on Computer Vision·2023
    Same journal

    The Heart of the Metaverse: How Immersive Technologies Are Revolutionizing Cardiac Care.

    IEEE pulse·2026
    Same journal

    Benefits for Early Diagnosis, Treatment, and Research.

    IEEE pulse·2026
    Same journal

    At the Crossroads of Innovation.

    IEEE pulse·2026
    Same journal

    Robotics in the Cath Lab: Precision, Safety, and the Rise of Remote Cardiac Interventions.

    IEEE pulse·2026
    Same journal

    Industry Corner Live With BioBeat CEO Arik Ben Ishay.

    IEEE pulse·2026
    Same journal

    Engineering the Next Generation of Artificial Hearts.

    IEEE pulse·2026
    See all related articles

    We developed MedProbCLIP, a new probabilistic model for chest X-ray and radiology reports. It significantly improves trustworthiness and performance in retrieval and classification tasks.

    Area of Science:

    • Biomedical informatics
    • Artificial intelligence in healthcare
    • Medical imaging analysis

    Background:

    • Vision-language foundation models offer powerful general-purpose learning but lack reliability for critical biomedical applications due to deterministic embeddings.
    • High-stakes medical tasks necessitate dependable AI models that can provide trustworthy representations.

    Purpose of the Study:

    • To introduce MedProbCLIP, a novel probabilistic vision-language learning framework.
    • To enhance representation learning and retrieval for chest X-ray images and associated radiology reports.
    • To improve the trustworthiness of AI models in medical imaging analysis.

    Main Methods:

    • Developed MedProbCLIP, a probabilistic framework for vision-language learning.
    • Utilized chest X-ray images and radiology reports for model training and evaluation.

    Related Experiment Videos

  • Employed the MIMIC-CXR dataset for comprehensive performance assessment.

    • Main Results:

    • MedProbCLIP demonstrated superior performance compared to deterministic and other probabilistic baselines.
    • Achieved significant improvements in both image-report retrieval tasks.
    • Showcased enhanced capabilities in zero-shot classification of medical data.
    • Markedly increased the trustworthiness of the learned representations.

    Conclusions:

    • MedProbCLIP offers a more reliable and trustworthy approach for vision-language learning in the biomedical domain.
    • The probabilistic framework addresses the limitations of deterministic models in high-stakes medical tasks.
    • This advancement holds promise for improving AI-driven diagnostic and retrieval systems in radiology.