Related Experiment Video
Updated: Oct 22, 2025
07:22
Surgical Robot-Assisted Transanal Specimen Extraction Radical Sigmoidectomy Without an Auxiliary Abdominal Incision
Published on: June 13, 2025
281
Endoscopic Image-Based Skill Assessment in Robot-Assisted Minimally Invasive Surgery
Gábor Lajkó1,2, Renáta Nagyné Elek3,4,5, Tamás Haidegger3,6
1Autonomous Systems Track, Double Degree Programme, EIT Digital Master School, Technische Universität Berlin, Straße des 17. Juni 135, 10623 Berlin, Germany.
Sensors (Basel, Switzerland)
|August 28, 2021
Summary
This study introduces a 2D image-based solution for objective surgical skill assessment in any training environment. The method achieved over 80% accuracy for knot-tying, needle-passing, and suturing tasks, enhancing surgical training feedback.
Area of Science:
- Medical training and simulation
- Computer vision in surgery
- Surgical skill assessment
Background:
- Objective skill assessment is crucial for surgical training, but kinematic data is unavailable for traditional manual Minimally Invasive Surgery (MIS).
- 2D visual features offer a viable alternative for skill assessment in MIS training environments, unlike limited 3D methods.
- Existing methods for skill assessment can be enhanced with additional sensors or advanced feature extraction techniques.
Purpose of the Study:
- To introduce a general 2D image-based solution for objective surgical skill assessment applicable to any training environment.
- To evaluate the accuracy of this 2D feature-based approach using the JHU-ISI Gesture and Skill Assessment Working Set (JIGSAWS) dataset.
- To compare the performance of different feature extraction techniques and classification methods for surgical skill assessment.
Main Methods:
- Utilized a 2D image-based approach for skill assessment, processing features using established extraction techniques.
- Employed the JHU-ISI Gesture and Skill Assessment Working Set (JIGSAWS) dataset for evaluation and comparative analysis.
- Assessed individual trial accuracy and mean accuracy across five cross-validation trials for surgical subtasks like Knot-Tying, Needle-Passing, and Suturing.
Main Results:
- The algorithm achieved up to 95.74% accuracy in individual trials.
- Mean accuracy for surgical subtasks: Knot-Tying (83.54%), Needle-Passing (84.23%), and Suturing (81.58%).
- The proposed 2D visual-based method demonstrated state-of-the-art performance, exceeding 80% accuracy for all evaluated JIGSAWS subtasks.
Conclusions:
- The developed 2D image-based solution provides an effective method for objective surgical skill assessment in diverse training settings.
- The approach shows significant potential for improving surgical training feedback by accurately evaluating surgeon performance.
- Further improvements in accuracy are possible through the integration of novel visual features and advanced classification algorithms.
