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

You might also read

Related Articles

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

Sort by
Same author

From ecological foundations to nature-based solutions: A semantic-temporal scientometric analysis of urban expansion and green space research.

Journal of environmental management·2026
Same author

Feline Assistance: Cats, unlike dogs and toddlers, help you only when it helps them.

Scientific American·2026
Same author

Patronus: patient-specific training setup for aortic aneurysm repair.

International journal of computer assisted radiology and surgery·2026
Same author

Real-time planar angiographic cerebral perfusion imaging.

International journal of computer assisted radiology and surgery·2026
Same author

Lighting the Way: The forest may be glowing-at least to deer.

Scientific American·2026
Same author

Imputing missing values in single-cell RNA-sequencing data: a statistical and machine learning-based approach.

Briefings in bioinformatics·2026

Related Experiment Video

Updated: May 22, 2025

Author Spotlight: Segmentation and VR for Advanced Neurovascular Interventions
06:18

Author Spotlight: Segmentation and VR for Advanced Neurovascular Interventions

Published on: April 5, 2024

925

SASVi: segment any surgical video.

Ssharvien Kumar Sivakumar1, Yannik Frisch2,3, Amin Ranem4

  • 1GRIS, TU Darmstadt, Fraunhoferstr. 5, 64283, Darmstadt, Germany. ssharvien_kumar.sivakumar@tu-darmstadt.de.

International Journal of Computer Assisted Radiology and Surgery
|May 20, 2025
PubMed
Summary
This summary is machine-generated.

We developed SASVi, a re-prompting method for surgical video segmentation using foundation models. SASVi improves temporal consistency by adapting to changing scenes, even with limited data.

Keywords:
Foundation modelsSurgical video segmentationTemporal consistency

More Related Videos

Robotized Testing of Camera Positions to Determine Ideal Configuration for Stereo 3D Visualization of Open-Heart Surgery
05:12

Robotized Testing of Camera Positions to Determine Ideal Configuration for Stereo 3D Visualization of Open-Heart Surgery

Published on: August 12, 2021

1.9K
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

2.6K

Related Experiment Videos

Last Updated: May 22, 2025

Author Spotlight: Segmentation and VR for Advanced Neurovascular Interventions
06:18

Author Spotlight: Segmentation and VR for Advanced Neurovascular Interventions

Published on: April 5, 2024

925
Robotized Testing of Camera Positions to Determine Ideal Configuration for Stereo 3D Visualization of Open-Heart Surgery
05:12

Robotized Testing of Camera Positions to Determine Ideal Configuration for Stereo 3D Visualization of Open-Heart Surgery

Published on: August 12, 2021

1.9K
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

2.6K

Area of Science:

  • Computer Vision
  • Medical Imaging
  • Artificial Intelligence

Background:

  • Foundation models require domain-specific adaptation for tasks like surgical video segmentation.
  • Existing methods struggle with dynamic scenes where objects enter or leave the view.

Purpose of the Study:

  • To develop a novel re-prompting mechanism for temporally consistent surgical video segmentation.
  • To enable foundation models to generalize to the specific semantics of surgical environments.

Main Methods:

  • Proposed SASVi, a re-prompting mechanism utilizing an Overseer object detection model.
  • Trained the Overseer model on minimal target domain annotations.
  • Automated re-prompting of the foundation model SAM2 based on scene changes.

Main Results:

  • SASVi improved temporal consistency in surgical video segmentation by at least 2.4% compared to frame-wise methods.
  • Demonstrated successful deployment of SAM2 on surgical videos across multiple datasets.
  • Achieved quantitative and qualitative improvements in segmentation accuracy.

Conclusions:

  • SASVi establishes a new baseline for temporally consistent surgical video segmentation with limited annotations.
  • The method effectively leverages scarce data to generate complete video annotations.
  • Publicly released annotations will support future surgical data science model development.