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 Concept Videos

Positron Emission Tomography01:29

Positron Emission Tomography

Positron emission tomography (PET) is a medical imaging technique involving radiopharmaceuticals — substances that emit short-lived radiation. Although the first PET scanner was introduced in 1961, it took 15 more years before radiopharmaceuticals were combined with the technique and revolutionized its potential.
One of the main requirements of a PET scan is a positron-emitting radioisotope, which is produced in a cyclotron and then attached to a substance used by the part of the body being...
Imaging Studies II: Positron Emission Tomography and Scintigraphy01:25

Imaging Studies II: Positron Emission Tomography and Scintigraphy

Positron Emission Tomography (PET) is a medical imaging technique that provides crucial insights into the body's physiological functions at a molecular level. It is an indispensable resource for diagnosing, staging, and monitoring various illnesses, notably cancer, neurological disorders, and cardiovascular conditions.
Fundamental Principles of PET

You might also read

Related Articles

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

Sort by
Same author

Resolving phenotyping discordance with SPACEMAP, an integrated machine learning framework.

bioRxiv : the preprint server for biology·2025
Same author

Universal consensus 3D segmentation of cells from 2D segmented stacks.

Nature methods·2025
Same author

Prompts to Table: Specification and Iterative Refinement for Clinical Information Extraction with Large Language Models.

medRxiv : the preprint server for health sciences·2025
Same author

Benchmarks of success in radiotherapy vs systemic therapy: National Clinical Trials Network (NCTN) randomized controlled trials sponsored by the National Cancer Institute (NCI).

Journal of the National Cancer Institute·2024
Same author

Universal consensus 3D segmentation of cells from 2D segmented stacks.

bioRxiv : the preprint server for biology·2024
Same author

A rapid and bidirectional reporter of neural activity reveals neural correlates of social behaviors in Drosophila.

Nature neuroscience·2023

Related Experiment Video

Updated: May 10, 2026

Application of Deep Learning-Based Medical Image Segmentation via Orbital Computed Tomography
04:48

Application of Deep Learning-Based Medical Image Segmentation via Orbital Computed Tomography

Published on: November 30, 2022

3.3K

Physical Examination Identification in Medical Education Videos: Zero-Shot Multimodal AI With Temporal Sequence

Shinyoung Kang1, Michael Holcomb1, David Hein1

  • 1Lyda Hill Department of Bioinformatics, The University of Texas Southwestern Medical Center, Dallas, TX, United States.

JMIR AI
|December 18, 2025
PubMed
Summary

Multimodal large language models (MM-LLMs) can automatically segment physical examination (PE) periods in Objective Structured Clinical Examination (OSCE) videos. This AI approach significantly reduces video review time for medical education assessments.

Keywords:
AIartificial intelligencemedical educationmultimodal large language modelsvideo segmentation

More Related Videos

Objectification of Tongue Diagnosis in Traditional Medicine, Data Analysis, and Study Application
05:56

Objectification of Tongue Diagnosis in Traditional Medicine, Data Analysis, and Study Application

Published on: April 14, 2023

3.1K
Author Spotlight: Addressing Technical and Subjective Challenges in Measuring Classroom Attention
06:37

Author Spotlight: Addressing Technical and Subjective Challenges in Measuring Classroom Attention

Published on: December 15, 2023

5.2K

Related Experiment Videos

Last Updated: May 10, 2026

Application of Deep Learning-Based Medical Image Segmentation via Orbital Computed Tomography
04:48

Application of Deep Learning-Based Medical Image Segmentation via Orbital Computed Tomography

Published on: November 30, 2022

3.3K
Objectification of Tongue Diagnosis in Traditional Medicine, Data Analysis, and Study Application
05:56

Objectification of Tongue Diagnosis in Traditional Medicine, Data Analysis, and Study Application

Published on: April 14, 2023

3.1K
Author Spotlight: Addressing Technical and Subjective Challenges in Measuring Classroom Attention
06:37

Author Spotlight: Addressing Technical and Subjective Challenges in Measuring Classroom Attention

Published on: December 15, 2023

5.2K

Area of Science:

  • Artificial Intelligence in Medical Education
  • Computer Vision for Clinical Assessment
  • Natural Language Processing in Healthcare

Background:

  • Objective Structured Clinical Examinations (OSCEs) are crucial for assessing medical student competency but are resource-intensive.
  • Current automated video analysis methods struggle with long medical videos due to computational and temporal context challenges.
  • The physical examination (PE) component is vital yet often a small, difficult-to-isolate part of OSCE recordings.

Purpose of the Study:

  • To evaluate the efficacy of multimodal large language models (MM-LLMs) in segmenting PE periods within OSCE videos.
  • To determine if MM-LLMs can reduce the evaluation burden without prior specific training.
  • To assess the potential of AI in streamlining the assessment of clinical skills.

Main Methods:

  • Analysis of 500 15-minute OSCE videos with hand-labeled PE periods as ground truth.
  • Frame sampling at 1-3 second intervals with pose detection preprocessing.
  • Utilized six MM-LLMs for frame-level classification and a hidden Markov model for temporal consistency.
  • Performance metrics included recall, precision, Intersection over Union (IOU), and predicted PE length.

Main Results:

  • GPT-4o achieved high recall (0.998) and IOU (0.784) at 1-second sampling, reducing review time by 81%.
  • The model identified 175 seconds of PE per video versus a mean labeled 126 seconds.
  • Other MM-LLMs showed trade-offs; GPT-4o provided the best balance of recall and precision.
  • Error analysis indicated areas for improvement in handling occluded maneuvers and preparatory actions.

Conclusions:

  • Zero-shot MM-LLMs combined with temporal modeling effectively segment PE in OSCE videos, minimizing the need for extensive training data.
  • This AI-driven approach significantly cuts review time while preserving clinical assessment integrity.
  • The technique offers a scalable foundation for efficient clinical skill assessment in medical education.