Related Experiment Video
Updated: Jan 8, 2026
04:48
Application of Deep Learning-Based Medical Image Segmentation via Orbital Computed Tomography
Published on: November 30, 2022
3.3K
Semantic segmentation deep learning model boosts surgeons' organ recognition in minimally invasive hysterectomy - a
Shin Takenaka1,2,3, Hiroki Matsuzaki4, Yusuke Hirose3
1Department of Gynecology, National Cancer Center Hospital East, Chiba, Japan.
International Journal of Surgery (London, England)
|December 22, 2025
Summary
Artificial intelligence (AI) enhances surgeon recognition of the ureter and bladder during minimally invasive hysterectomy. This AI support improves detection rates, especially for less experienced surgeons, without compromising accuracy.
Area of Science:
- Surgical innovation
- Medical imaging analysis
- Artificial intelligence in healthcare
Background:
- Minimally invasive hysterectomy carries risks of ureter and bladder injury due to inadequate intraoperative organ recognition.
- Artificial intelligence (AI) offers potential solutions for improving anatomical identification and mitigating surgical risks.
Purpose of the Study:
- To evaluate an AI-based anatomical recognition system for ureter and bladder identification during minimally invasive hysterectomy.
- To assess the impact of AI support on surgeon organ recognition abilities across experience levels.
- To determine if AI assistance compromises specificity in organ identification.
Main Methods:
- A deep learning model was trained on a large dataset of ureter (13,934 images) and bladder (4,940 images) from 41 institutions.
- Model performance was quantified using the Dice coefficient.
- The impact of AI on surgeon performance (sensitivity and specificity) was assessed in 16 surgeons across eight facilities using pre-recorded surgical videos.
Main Results:
- The AI model achieved Dice coefficients of 0.66 for ureter and 0.62 for bladder segmentation.
- AI assistance significantly improved ureter detection sensitivity (43.5% to 58.1%) and bladder detection sensitivity (54.2% to 70.0%), with p < 0.001 for both.
- The benefits of AI were more pronounced in less experienced surgeons, showing sensitivity improvements of 27.3% for ureter and 26.8% for bladder recognition, without affecting specificity.
Conclusions:
- AI-based systems demonstrably improve surgeons' ability to recognize ureters and bladders in surgical videos.
- The AI system enhances recognition particularly for less experienced surgeons, without increasing misidentification risks.
- AI holds promise for improving intraoperative anatomical identification, leading to safer and more standardized surgical procedures.
