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

Surg-NAT+: negation-aware vision-language refinement for fine-grained surgical understanding.

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

GEN-Guard: correcting generalization failures for deployable federated surgical AI.

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

Endoshare: a publicly available, surgeons-friendly solution to de-identify and manage surgical videos.

Surgical endoscopy·2026
Same author

High-intensity focused ultrasound (HIFU) modeling: in vitro validation and integration into patient-specific planning tool.

Scientific reports·2026
Same author

SurgViVQA: temporally grounded video question answering for surgical scene understanding.

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

S4M: 4-points to segment anything.

International journal of computer assisted radiology and surgery·2026

Related Experiment Video

Updated: May 31, 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

951

SAF-IS: A spatial annotation free framework for instance segmentation of surgical tools.

Luca Sestini1, Benoit Rosa2, Elena De Momi3

  • 1ICube, University of Strasbourg, CNRS, France; Department of Electronics, Information and Bioengineering, Politecnico di Milano, Milano, Italy.

Medical Image Analysis
|January 24, 2025
PubMed
Summary

This study introduces a new framework for surgical instrument instance segmentation that avoids costly pixel-level annotations. It uses binary masks and tool presence labels for training, achieving state-of-the-art results without spatial data.

Keywords:
Endoscopic videosInstance segmentationSelf supervised learningTool segmentationWeakly supervised learning

More Related Videos

A Comprehensive Protocol for Manual Segmentation of the Medial Temporal Lobe Structures
12:30

A Comprehensive Protocol for Manual Segmentation of the Medial Temporal Lobe Structures

Published on: July 2, 2014

20.3K
From Voxels to Knowledge: A Practical Guide to the Segmentation of Complex Electron Microscopy 3D-Data
12:08

From Voxels to Knowledge: A Practical Guide to the Segmentation of Complex Electron Microscopy 3D-Data

Published on: August 13, 2014

24.5K

Related Experiment Videos

Last Updated: May 31, 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

951
A Comprehensive Protocol for Manual Segmentation of the Medial Temporal Lobe Structures
12:30

A Comprehensive Protocol for Manual Segmentation of the Medial Temporal Lobe Structures

Published on: July 2, 2014

20.3K
From Voxels to Knowledge: A Practical Guide to the Segmentation of Complex Electron Microscopy 3D-Data
12:08

From Voxels to Knowledge: A Practical Guide to the Segmentation of Complex Electron Microscopy 3D-Data

Published on: August 13, 2014

24.5K

Area of Science:

  • Computer-assisted surgery
  • Medical image analysis
  • Deep learning for surgical robotics

Background:

  • Instance segmentation of surgical instruments is critical for computer-assisted surgery applications.
  • Current methods rely on fully-supervised deep learning models requiring expensive pixel-level annotations.
  • This annotation bottleneck hinders the widespread adoption of advanced surgical tools.

Purpose of the Study:

  • To develop a novel framework for surgical instrument instance segmentation that eliminates the need for spatial annotations during training.
  • To leverage readily available data like binary tool masks and tool presence labels for effective model training.
  • To enable more accessible and scalable development of computer-assisted surgery systems.

Main Methods:

  • A framework is proposed that utilizes binary tool masks and tool presence labels for training.
  • The method learns to extract individual tool instances and encode them into compact vector representations.
  • A small subset of instances is selected for human operator labeling, guiding the training of a tool instance classifier.

Main Results:

  • The framework was validated on the EndoVis 2017 and 2018 segmentation datasets.
  • Results were demonstrated using both manually annotated and unsupervised predicted binary masks.
  • The approach using unsupervised binary masks achieved instance segmentation completely free from spatial annotations.
  • Performance surpassed several state-of-the-art fully-supervised segmentation methods.

Conclusions:

  • The developed framework offers an effective solution for surgical instrument instance segmentation without requiring expensive pixel-level annotations.
  • Utilizing unsupervised binary segmentation models for mask generation results in a fully annotation-free approach.
  • This method significantly advances the feasibility of developing advanced computer-assisted surgery applications by reducing data acquisition costs.