A Study on the Effectiveness of AI-assisted Colonoscopy in Improving the Effect of Colonoscopy Training for Trainees

Wuhan University

Status: Unknown

Conditions: Gastrointestinal Disease; Artificial Intelligence; Colonoscopy

Treatments: Device: artificial intelligence assistance system

Study type: Interventional

Funder types: Other

Identifiers

NCT04912037

EA-21-005

Details and patient eligibility

About

In this study,the AI-assisted system（EndoAngel）has the functions of reminding the ileocecal junction, withdrawal time, withdrawal speed, sliding lens, polyps in the field of vision, etc. These functions can improve the colonoscopy performance of novice physicians and assist the colonoscopy training。

Full description

Colonoscopy is a key technique for detecting and diagnosing lesions of the lower digestive tract.High-quality endoscopy leads to better disease outcomes.However, the demand for endoscopy is high in China, and endoscopy is in short supply.A colonoscopy is a complex technical procedure that requires training and experience for maximal accuracy and safety.Therefore, it is of great significance to improve the colonoscopy ability of novice physicians and shorten the colonoscopy training time for solving the problems such as the lack and uneven distribution of digestive endoscopists and the substandard quality of endoscopy in China.

In recent years, deep learning algorithms have been continuously developed and increasingly mature.They have been gradually applied to the medical field. Computer vision is a science that studies how to make machines "see". Through deep learning, camera and computer can replace human eyes to carry out machine vision such as target recognition, tracking and measurement.Interdisciplinary cooperation in the field of medical imaging and computer vision is also one of the research hotspots in recent years. At present, it is mainly applied to the automatic identification and detection of lesions and quality control, and has achieved good results.

Our preliminary experiments have shown that deep learning has a high accuracy in endoscopic quality monitoring, which can effectively regulate doctors' operations, reduce blind spots and improve the quality of endoscopic examination.At the same time, it can also monitor the doctor's withdrawal time in real time and improve the detection rate of adenoma.In the previous work of our research group, we have successfully developed deep learning-based colonoscopy withdraw speed monitoring and intestinal cleanliness assessment, and verified the effectiveness of the AI-assisted system（EndoAngel） in improving the quality of gastroscopy and colonoscopy in clinical trials.

Based on the above rich foundation of preliminary work, as well as the huge demand in the field of colonoscopy training,By comparing the colonoscopy operation training for novices with and without EndoAngel assistance, we plan to compare the colonoscopy learning effect of novices with and without assistance, including skill results and cognitive level, to explore whether AI can promote the improvement of the colonoscopy operation training for novices.

Enrollment: 385 estimated patients

Sex: All

Ages: 50+ years old

Volunteers: No Healthy Volunteers

Inclusion criteria

1. Male or female ≥50 years old;
2. Able to read, understand and sign informed consent
3. The investigator believes that the subjects can understand the process of the clinical study, are willing and able to complete all study procedures and follow-up visits, and cooperate with the study procedures
4. Patients requiring colonoscopy

Exclusion criteria

1. Have drug or alcohol abuse or mental disorder in the last 5 years
2. Pregnant or lactating women
3. Patients with known multiple polyp syndrome;
4. patients with known inflammatory bowel disease;
5. known intestinal stenosis or space-occupying tumor;
6. known colon obstruction or perforation;
7. patients with a history of colorectal surgery;
8. Patients with previous history of allergy to pre-used spasmolysis;
9. Unable to perform biopsy and polyp removal due to coagulation disorders or oral anticoagulants;
10. High risk diseases or other special conditions that the investigator considers the subject unsuitable for participation in the clinical trial.

Trial design

Primary purpose: Diagnostic

Allocation: Randomized

Interventional model: Parallel Assignment

Masking: Double Blind

---

385 participants in 2 patient groups

with AI-assisted system

Experimental group

Description:

The novice doctors are trained in colonoscopy with an artificial intelligence assisted system that can indicate abnormal lesions and the speed of withdrawal in real time, as well as feedback on the percentage of overspeed.

Treatment:

Device: artificial intelligence assistance system

without AI-assisted system

No Intervention group

Description:

The novice doctors receive routine colonoscopy training without artificial intelligence assistance system and no special tips