Effectiveness of an Ai-based Endoscopic Ultrasound Navigation System in the Training of Endoscopic Ultrasonics

Feasibility and Effectiveness of an Ai-based Endoscopic Ultrasound Navigation System in the Training of Endoscopic ultrasonics-a Prospective, Randomized, Multi-center Clinical Trial

In the early stage, the investigators successfully constructed an artificial intelligence model-based ultrasonic endoscopy-assisted film reading system and named the modified system biliopancreatic Master. The system can realize real-time ultrasonic station recognition and anatomical mark recognition and provide doctors with corresponding operation techniques. This study aimed to verify the feasibility and effectiveness of the biliopancreatic master system developed by our project team in shortening the training period of ultrasound endoscopists through a single-center clinical study.

Study Overview

• Status: Not yet recruiting
• Conditions: Endoscopic Ultrasound
• Intervention / Treatment: Device: Artificial intelligence assisted system

Detailed Description

Endoscopic ultrasonography is essential for diagnosing and treating biliary and pancreatic diseases. The total incidence of biliary and pancreatic diseases is as high as 121/100,000, causing about 14.3/100,000 deaths worldwide every year, posing a severe threat to the health and safety of all humanity. The biliopancreatic system is located deep in the abdominal cavity, which is relatively difficult to examine and lacks specific symptoms at the early stage of the disease. Hence, the prognosis is poor, and the five-year survival rate is low. Biliopancreatic ultrasound endoscopy can obtain the cross cross-section of the biliopancreatic system through ultrasonic detection, collect tissue samples for pathological biopsy through the puncture, and carry out operations such as flushing and drainage, radiofrequency ablation, abdominal ganglion block, and gastrojejunostomy. The device integrates detection and treatment. In terms of detection, endoscopic ultrasonography is close to the lesion tissue and can avoid influencing the abdominal cavity wall, gastrointestinal gas, and other organs. Compared with X-ray, CT, MRI, and different in vitro detection methods, endoscopic ultrasonography has technical advantages of high resolution, no ionizing radiation, and solid real-time performance. In terms of treatment, endoscopic ultrasound-guided puncture through natural lumen has the characteristics of minimally invasive, lower surgical risk and cost compared with surgical treatment, which can significantly improve the prognosis of patients with biliary and pancreatic diseases. Biliopancreatic endoscopic ultrasonography is essential for diagnosing and treating biliopancreatic system diseases.

The basis of endoscopic ultrasound diagnosis and treatment lies in the accurate identification and localization of lesions by ultrasound images. Ultrasound images are cross-section images of human tissues, mainly containing texture information that is difficult to be recognized by naked human eyes; for endoscopy, doctors lacking professional training and long-term practice, difficult to accurately identify the anatomical signs in the images, which significantly affects the accuracy of identification and localization of biliopancreatic lesions. Due to the low image recognition of ultrasonic endoscopy, the operation of ultrasonic endoscopy is difficult, and the training cycle is very long. Conventional European and American endoscopic ultrasound training programs lasted for three years, but in a clinical study, two-thirds of the physicians who participated in the standard training program failed to pass the ability assessment of endoscopic ultrasound. Although ultrasound endoscopy is one of the most sensitive methods for the diagnosis of pancreatic cancer, sensitivity (87.8%) is still significantly inadequate compared with optical endoscopy for the diagnosis of gastric cancer (96%) and colon cancer (97.1%). The existence of this problem means that the existing training in ultrasonic endoscopy needs to be optimized. In recent years, Artificial intelligence (AI) machine vision technology has developed rapidly and has been widely used in transportation, finance, medicine and other fields. Several studies have shown that the diagnostic ability of AI has surpassed that of human experts in some diseases, and some studies have confirmed the feasibility of the application of AI in endoscopic ultrasound image recognition. In a previous study, an artificial intelligence model was used to assist ultrasonic endoscopists in ultrasound film reading, which could significantly improve the accuracy of physicians in image positioning and anatomical marker segmentation. In the early stage, the investigators successfully constructed an artificial intelligence model-based ultrasonic endoscopy-assisted film reading system and named the modified system biliopancreatic Master. The system can realize real-time ultrasonic station recognition and anatomical mark recognition and provide doctors with corresponding operation techniques.

• Study Type: Interventional
• Enrollment (Estimated): 12
• Phase: Not Applicable

Contacts and Locations

• Study Contact: Name: Yu Honggang, Doctor; Phone Number: 13871281899; Email: yuhonggang@whu.edu.cn
• Study Contact Backup: Name: : Yu Honggang, Doctor; Phone Number: : 13871281899; Email: yuhonggang@whu.edu.cn

Study Locations

• China
  • Guangdong
    • Shenzhen, Guangdong, China, 518066
      • The Eighth Affiliated Hospital, Sun Yat-Sen University
      • Contact: Honglei Chen, Doctor; Email: chenhlei3@mail.sysu.edu.cn
  • Hubei
    • Wuhan, Hubei, China, 430060
      • Renmin Hospital of Wuhan University

Participation Criteria

Eligibility Criteria

• Ages Eligible for Study: 18 years and older (Adult, Older Adult)
• Accepts Healthy Volunteers: No

Description

Endoscopist:

Inclusion Criteria:

Endoscopists with experience in performing more than 300 gastroscopies; Lack of experience in independently operating an endoscopic ultrasound (EUS); No experience in EUS-guided fine-needle aspiration biopsy (FNA).

Patient:

Inclusion Criteria:

Patient age ≥18 years; Patients who consecutively undergo sedated EUS procedures; Ability to read, understand, and sign the informed consent; Patients suspected of having biliary (both pancreatic and biliary) lesions based on clinical symptoms and/or radiological findings and/or laboratory test results; High-risk patients for pancreatic cancer: known genetic mutations associated with the risk of pancreatic cancer (BRCA2, BRCA1, PALB2, ATM, CDKNA/p16); familial pancreatic ductal adenocarcinoma with no known lineage mutations; Peutz-Jeghers syndrome (STK11); Lynch syndrome (MLH1/MSH2/MSH6, EPCAM, PMS2); familial adenomatous polyposis (APC) etc. Based on the preoperative cholangiopancreatography report, patients are classified into those with or without radiological findings. Patients with radiological findings are categorized by lesion location, such as the pancreas, PD, CBD, and other lesions (like gallbladder, duodenal papilla suspected of biliary invasion).

Exclusion Criteria:

Patients with absolute contraindications to EUS examination; Previous gastric surgery; Pregnancy; Severe internal medical diseases; History of allergy to anesthesia drugs; Esophageal narrowing or obstruction; Upper gastrointestinal anatomical abnormalities caused by advanced tumors.

Study Plan

How is the study designed?

Design Details

• Primary Purpose: Health Services Research
• Allocation: Randomized
• Interventional Model: Parallel Assignment
• Masking: None (Open Label)

Number of Arms

2

Arms and Interventions

Participant Group / Arm	Intervention / Treatment
Experimental: A novice with ai-assisted assistance; When patients are randomly assigned to the experimental group, the visiting physician will perform endoscopic ultrasound with the assistance of the ai-assisted system	Device: Artificial intelligence assisted system; A deep learning-based bile duct scanning system that can prompt endoscopists to scan standard stations, identify bile ducts and stones in real time
No Intervention: A novice without ai-assisted assistance; When the patients were randomly assigned to the control group, the visiting physician would not perform endoscopic ultrasonography with the assistance of the AI-assisted system

What is the study measuring?

Primary Outcome Measures

Outcome Measure	Measure Description	Time Frame
Learning cycle for ultrasound endoscopists	The number of guided examinations required by the trainees to achieve the level of stable independent operation of ultrasonic endoscopy	No more than 1 year

Secondary Outcome Measures

Outcome Measure	Measure Description	Time Frame
Duration of operation by ultrasound endoscopists	The time required to complete each operation, from the beginning to the end of the operation	No more than 1 year

Collaborators and Investigators

• Sponsor: Renmin Hospital of Wuhan University
• Investigators: Principal Investigator: Yu Honggang, Doctor, Renmin Hospital of Wuhan University

Study record dates

Study Major Dates

• Study Start (Estimated): October 8, 2023
• Primary Completion (Estimated): May 31, 2024
• Study Completion (Estimated): May 31, 2024

Study Registration Dates

• First Submitted: May 21, 2022
• First Submitted That Met QC Criteria: June 12, 2022
• First Posted (Actual): June 16, 2022

Study Record Updates

• Last Update Posted (Actual): August 24, 2023
• Last Update Submitted That Met QC Criteria: August 22, 2023
• Last Verified: May 1, 2023

More Information

Terms related to this study

• Keywords: Artificial intelligence; Learning curve; Ultrasound Endoscopy
• Other Study ID Numbers: EA-22-005

Plan for Individual participant data (IPD)

• Plan to Share Individual Participant Data (IPD)?: NO

Drug and device information, study documents

• Studies a U.S. FDA-regulated drug product: No
• Studies a U.S. FDA-regulated device product: No