Related Experiment Video
Updated: Aug 30, 2025
Laparoscopic Anatomical Hepatectomy Using Takasaki's Approach and Indocyanine Green Fluorescence Navigation
Published on: May 16, 2025
Automated surgical workflow identification by artificial intelligence in laparoscopic hepatectomy: Experimental
Kimimasa Sasaki1, Masaaki Ito2, Shin Kobayashi3
1Surgical Device Innovation Office, National Cancer Center Hospital East, 6-5-1, Kashiwanoha, Kashiwa-City, Chiba, 277-8577, Japan; Department of Hepatobiliary and Pancreatic Surgery, National Cancer Center Hospital East, 6-5-1, Kashiwanoha, Kashiwa-City, Chiba, 277-8577, Japan; Course of Advanced Clinical Research of Cancer, Juntendo University Graduate School of Medicine, 2-1-1, Hongo, Bunkyo-Ward, Tokyo, 113-8421, Japan.
A new deep-learning model accurately identifies surgical steps during laparoscopic hepatectomy (LH). This automated system enhances computer-assisted surgery (CAS) for safer procedures.
Area of Science:
- Surgical robotics and artificial intelligence
- Medical imaging and computer vision
- Gastrointestinal surgery
Background:
- Accurate laparoscopic hepatectomy (LH) requires advanced intraoperative tools.
- Computer-assisted surgery (CAS) systems are crucial for improving LH precision.
- Automated surgical workflow identification is a key component for developing CAS.
Purpose of the Study:
- To develop a deep-learning model for automated surgical step identification in LH.
- To enhance the accuracy and safety of LH procedures through AI.
Main Methods:
- A dataset of 40 pure LH videos was created, with 30 for training and 10 for testing.
- Videos were processed into frames, with surgical steps annotated.
- Two deep-learning models (8-step and 6-step) were developed using convolutional neural networks.
Main Results:
- Over 8 million frames were annotated for surgical step identification.
- Model 1 achieved 0.891 overall accuracy, improved to 0.947 in Model 2.
- Model 2 demonstrated high median (0.927) and average (0.937) accuracy in real-time identification at 21 frames per second.
Conclusions:
- A highly accurate deep-learning model for surgical step identification in pure LH was developed.
- The model shows potential for integration into intraoperative CAS systems.
- This advancement can contribute to safer and more precise laparoscopic surgeries.
