Research on Identifying Critical Surgical Anatomy in Cholecystectomy Videos Based on Deep Learning
Trial statusRecruiting
Trial phaseNot listed
Trial typeObservational
Biological sexAll
Age18+
SponsorChinese Academy of Sciences
About this trial
Laparoscopic cholecystectomy is a common surgical procedure, but it carries the potential for bile duct injury and other surgical risks. To provide visual assistance to surgeons during surgery and mitigate these risks, this research project aims to develop a real-time object recognition algorithm based on deep learning technology. This algorithm will label key anatomical structures in laparoscopic cholecystectomy videos, providing surgeons with immediate information on dangerous and safe areas.
Eligibility criteria
This trial does not accept healthy volunteersQualifiers
Patients aged 18 or above who are diagnosed by a doctor as needing laparoscopic cholecystectomy
Disqualifiers
Patients who did not undergo surgery at the original hospital and those whose videos were blurry were excluded.
Trial population
Patients aged 18 or above who are diagnosed by a doctor as needing laparoscopic cholecystectomy
Trial design
Design model
Cohort
Time perspective
Prospective
Treatments tested in this trial
AI-assisted Intraoperative Anatomy Analysis
Diagnostic testThis is a prospective study on patients aged 18 years or more diagnosed with laparoscopic cholecystectomy. We will collect information such as laparoscopic cholecystectomy videos and procedure type, excluding patients who did not undergo surgery at the original hospital or whose videos were blurry.
Treatment groups
200 Participants
are divided into 4 treatment groupsGroup A: The First Affiliated Hospital of Zhengzhou University1 intervention
Group B: Beijing Luhe Hospital, Capital Medical University1 intervention
Group C: Shanghai East Hospital of Tongji University1 intervention
Group D: Peking university people's hospital1 intervention
Trial outcomes
Primary outcomes
1
Dice Similarity Coefficient
Dice Similarity Coefficient is a statistical measure of the similarity between two sets of data. In the context of image segmentation, it is used to quantify the spatial overlap between a predicted segmentation mask and its corresponding ground truth mask.
Time frame
3 years2
Mean Intersection over Union
Mean Intersection over Union provides a measure of the overlap between the predicted segmentation and the ground truth, averaged across all classes present in the dataset.
Time frame
3 years3
Global Accuracy
The proportion of correctly classified pixels out of the total number of pixels in the image.
Time frame
3 yearsSecondary outcomes
1
Inference Latency
time taken by the algorithm to process a single video frame and generate the segmentation masks (inference latency), or equivalently, the number of frames processed per second
Time frame
3 yearsOther outcomes
1
Class-Specific Precision and Recall for Critical Structures
Precision: The proportion of predicted pixels for a structure that are actually part of that structure Recall: The proportion of actual ground truth pixels for a structure that were correctly identified by the algorithm
Time frame
3 yearsSponsors and contacts
Click on the lead sponsor to view all of their trials.
Chinese Academy of Sciences
Lead sponsor
The First Affiliated Hospital of Zhengzhou University
Collaborator
Capital Medical University
Collaborator
Beijing Anzhen Hospital
Collaborator
Shanghai East Hospital of Tongji University
Collaborator
Peking University People's Hospital
Collaborator
Source ClinicalTrials.gov