Related Experiment Video
Updated: Jun 12, 2025
11:19
Robotic Myotomy and Partial Fundoplication for Achalasia
Published on: August 11, 2023
1.4K
Artificial intelligence based surgical support for experimental laparoscopic Nissen fundoplication
Holger Till1, Ciro Esposito2, Chung Kwong Yeung3
1Department of Pediatric and Adolescent Surgery, Medical University of Graz, Graz, Austria.
Frontiers in Pediatrics
|June 9, 2025
Summary
This study developed an AI/CV model to classify laparoscopic Nissen fundoplication quality. The model accurately detects visible quality indicators, showing potential for real-time surgical support.
Area of Science:
- Artificial intelligence
- Computer vision
- Surgical technology
Background:
- AI/CV shows promise in medical imaging but lags in surgical applications.
- This study focuses on developing AI for laparoscopic surgery quality assessment.
Purpose of the Study:
- To develop the first image-based AI/CV model for classifying quality indicators of laparoscopic Nissen fundoplication (LNF).
- To evaluate the model's performance in distinguishing correct from incorrect surgical configurations.
Main Methods:
- Six visible quality indicators (VQIs) were defined for Nissen fundoplication.
- A porcine model was used, generating 57 video sequences and 3,138 images for annotation.
- Deep learning (EfficientNet) was employed to train image classifiers to detect VQIs.
Main Results:
- AI/CV models demonstrated strong performance in predicting VQIs, with an average F1-Score of 0.9738.
- A multi-class classifier showed similar performance, robust to image augmentation.
- Detection of incomplete and too loose wraps showed a slight decline in predictive power.
Conclusions:
- AI/CV algorithms can effectively detect VQIs in laparoscopic Nissen fundoplication images.
- This proof of concept provides experimental evidence for AI/CV in classifying surgical images.
- Future development could lead to AI-based real-time surgical support to improve outcomes and safety.
