Automatic Classification of Colorectal Polyps Using Probe-based Endomicroscopy With Artificial Intelligence

Shandong University200 enrolled

Overview

Probe-based confocal laser endomicroscopy (pCLE) is an endoscopic technique that enables real-time histological evaluation of gastrointestinal mucosa during ongoing endoscopy examination. It can predict the classification of Colorectal Polyps accurately. However this requires much experience, which limits the application of pCLE. The investigators designed a computer program using deep neural networks to differentiate hyperplastic from neoplastic polyps automatically in pCLE examination.

Study Type

INTERVENTIONAL

Allocation

RANDOMIZED

Purpose

DIAGNOSTIC

Masking

TRIPLE

Enrollment

200

Conditions

Probe-based Confocal Laser Endomicroscopy Artificial Intelligence Colorectal Polyps

Interventions

Eligibility

Sex: ALLMin age: 18 YearsMax age: 80 YearsHealthy volunteers:

Medical Language ↔ Plain English

Inclusion Criteria: aged between 18 and 80; agree to give written informed consent. Exclusion Criteria: Patients under conditions unsuitable for performing CLE including coagulopathy , impaired renal or hepatic function, pregnancy or breastfeeding, and known allergy to fluorescein sodium; Inability to provide informed consent

Outcomes

Primary Outcomes

The accuracy of classifying colorectal Polyps using Probe-based endomicroscopy with deep neural networks

The primary outcome is to test the diagnostic accuracy, sensitivity, specificity, PPV, NPV of the Artificial Intelligence for diagnosing Colorectal Polyps on real-time pCLE examination.

Time frame: 4 months

Secondary Outcomes

Contrast the diagnosis efficiency of Artificial Intelligence with endoscopists

The secondary outcome is to compare the diagnosis efficiency (including diagnostic accuracy, sensitivity, specificity, PPV, NPV for diagnosing Colorectal Polyps on real-time pCLE examination) between Artificial Intelligence and endoscopists.

Time frame: 3 month