Clinical Research on a Novel Deep-learning Based System in Mediastinal Endoscopic Ultrasound Scanning

Clinical Research on Navigation and Quality Control System of Mediastinal Ultrasound Endoscopy Based on Deep Learning

The goal of this clinical trial is to develop and verify the auxiliary role of the artificial intelligence system in mediastinal ultrasound endoscopic scanning. The main questions it aims to answer are as follows: 1.The comparison of the image recognition accuracy between the artificial intelligence system and the ultrasound endoscopist; 2. Whether the artificial intelligence system can improve the efficiency of the mediastinum scanning for the ultrasound endoscopist. Participants will undergo mediastinal EUS with or without the assistance of the artificial intelligence system.

Study Overview

• Status: Recruiting
• Conditions: Mediastinum Disease
• Intervention / Treatment: Device: AI system

Detailed Description

In this study, a total of 200 cases of mediastinal endoscopic ultrasound scanning videos will be collected. First of all, an artificial intelligence system based on deep learning for the navigation and quality control of mediastinal endoscopic ultrasonography will be established. Secondly, the artificial intelligence system will be used to identify the site and anatomical structure of the mediastinal ultrasound endoscope, and the results of the artificial intelligence system's station recognition will be compared with the results of the endoscopist's station recognition. Finally, compare the image recognition speed, image recognition accuracy and precision of endoscopists with and without the assistance of artificial intelligence system.

• Study Type: Interventional
• Enrollment (Anticipated): 200
• Phase: Not Applicable

Contacts and Locations

• Study Contact: Name: Xiaoyan Wang, Doctor; Phone Number: +8613974889301; Email: wxy20011@163.com
• Study Contact Backup: Name: Shiqin Huang, MD; Phone Number: +8618308312098; Email: sqhuang0213@163.com

Study Locations

• China
  • Hunan
    • Changsha, Hunan, China, 410013
      • Recruiting
      • The Third Xiangya Hospital of Central South University
      • Contact: Xiaoyan Wang, Doctor; Phone Number: +8613974889301; Email: wxy20011@163.com
      • Contact: Shiqin Huang, Doctor; Phone Number: +8618308312098; Email: sqhuang0213@163.com

Participation Criteria

Eligibility Criteria

• Ages Eligible for Study: 18 years to 80 years (Adult, Older Adult)
• Accepts Healthy Volunteers: Yes
• Genders Eligible for Study: All

Description

Inclusion Criteria:

• 1. Age ≥18 years old, <80 years old 2.Patients who need endoscopic ultrasonography of mediastinum; 3. Agree to participate in this study and sign the informed consent form.

Exclusion Criteria:

• Subjects who meet any of the following criteria cannot be selected for this trial:

First. The patient's physical condition does not meet the requirements of conventional endoscopic ultrasonography:

1. Poor physical condition, including hemoglobin ≤8.0g/dl, severe cardiopulmonary insufficiency, etc.
2. Anesthesia assessment failed
3. Pregnancy or breastfeeding
4. In the acute stage of chemical and corrosive injury, it is very easy to cause perforation
5. Recent acute coronary syndrome or clinically unstable ischemic heart attack
6. Heart disease patients with right-to-left shunt, patients with severe pulmonary hypertension (pulmonary artery pressure> 90mmHg),patients with uncontrolled systemic hypertension and patients with adult respiratory distress syndrome.

Second. Disagree to participate in this study.

Third. There are other problems that do not meet the requirements of this research or that affect the results of the research:

1. Mediastinal lesions have previously undergone surgery or radiotherapy and chemotherapy;
2. Mental illness, drug addiction, inability to express themselves or other diseases that may affect follow-up.

Study Plan

How is the study designed?

Design Details

• Primary Purpose: Other
• Allocation: Randomized
• Interventional Model: Parallel Assignment
• Masking: Double

Number of Arms

2

Arms and Interventions

Participant Group / Arm	Intervention / Treatment
Experimental: AI-assisted group; Subjects will undergo EUS examination with the assistance of AI system.	Device: AI system; Patients will undergo EUS examination with the assistance of AI system.
No Intervention: non-assisted group; Subjects will undergo EUS examination without the assistance of AI system

What is the study measuring?

Primary Outcome Measures

Outcome Measure	Measure Description	Time Frame
Accuracy	The number of correctly classified images divided by the total number of images.	1 year
intersection over union (IoU)	It was defined as the relative area of overlap between the predicted bounding box(A) and the ground-truth(B) bounding box.	1 year
Recall	Recall=TP/(TP+FN)	1 year
Dice	Dice=2TP/(2TP+FP+FN)	1 year
Precision	When the IoU was greater than the threshold, the prediction was true positive(TP); when the IoU was less than the threshold, the prediction was false positive(FP).When the model segmentation area was equal to 0, it was false negative(FN).Precision=TP/(TP+FP)	1 year

Secondary Outcome Measures

Outcome Measure	Measure Description	Time Frame
Cohen's kappa coefficient	This data is to evaluate the agreement between the model and the endoscopists.	1 year

Collaborators and Investigators

• Sponsor: The Third Xiangya Hospital of Central South University

Study record dates

Study Major Dates

• Study Start (Actual): December 1, 2022
• Primary Completion (Anticipated): December 31, 2023
• Study Completion (Anticipated): January 31, 2024

Study Registration Dates

• First Submitted: March 17, 2023
• First Submitted That Met QC Criteria: March 29, 2023
• First Posted (Actual): March 31, 2023

Study Record Updates

• Last Update Posted (Actual): March 31, 2023
• Last Update Submitted That Met QC Criteria: March 29, 2023
• Last Verified: March 1, 2023

More Information

Terms related to this study

• Keywords: training; EUS; AI; Quality Control
• Additional Relevant MeSH Terms: Respiratory Tract Diseases; Thoracic Diseases; Mediastinal Diseases
• Other Study ID Numbers: 2023-EUS-AI-002

Plan for Individual participant data (IPD)

• Plan to Share Individual Participant Data (IPD)?: NO
• IPD Plan Description: Some data underlying this trial cannot be shared publicly due to protection of participant's privacy.

Drug and device information, study documents

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