Real-time Artificial Intelligence-based Endocytoscopic Diagnosis of Colorectal Neoplasms

Real-time Artificial Intelligence-based Endocytoscopic Diagnosis of Colorectal Neoplasms: a Single Center, Prospective Clinical Study

Colorectal cancer (CRC) is the third most common malignancy and the second leading cause of cancer-related death worldwide. Colonoscopy is considered the preferred method of screening for colorectal cancer, and resection of colorectal lesions can significantly reduce the incidence and mortality of colorectal cancer. In order to improve the qualitative and quantitative diagnosis of colorectal lesions, many endoscopic techniques, such as image-enhanced endoscopy (IEE), including narrowband imaging (NBI), magnifying endoscopy, pigment endoscopy, confocal laser endoscopy, and endocytoscopy (EC) are applied clinically. However, with the increasing number of endoscopic resection, the costs associated with the pathological diagnosis of endoscopic resection and resection specimens increase year by year. In clinical practice, some non-neoplastic colorectal lesions may not require resection, so it is important to distinguish neoplastic from non-neoplastic during colonoscopy. The application of EC is intended to achieve the purpose of real-time histopathological endoscopic diagnosis without biopsy. Several studies have shown that EC is effective in identifying the nature of colorectal lesions and judging the depth of invasion in CRC. Based on the endoscopic diagnosis, the endoscopist can determine the treatment plan for the colorectal lesions. The latest EC is an integrated endoscope with a contact light microscopy system with a maximum magnification of 520 x. EC can demonstrate the atypical of gland structure and cells after staining and display the super-amplified surface microvessels of the lesion under the EC-NBI mode. However, the judgment of endocytoscopic images needs a lot of experience to improve the diagnostic accuracy. Moreover, endoscopists have certain subjective judgments and errors in endocytoscopic diagnosis. There is an artificial intelligence system which has been developed to identify colorectal neoplasms. However, there is still a lack of prospective clinical verification based on Chinese population. In the study, the investigators performed a prospective clinical study to determine the diagnostic accuracy of artificial intelligence system.

Study Overview

• Status: Completed
• Conditions: Colorectal Neoplasms
• Intervention / Treatment: Diagnostic test: artificial intelligence system

Detailed Description

Colonoscopy is currently the gold standard of screening for CRC. The endocytoscopy, due to its high magnification function, can achieve the purpose of optical biopsy. However, endoscopic doctors have certain difficulties in diagnosing with the endocytoscopy, especially for novice endoscopic doctors, whose diagnostic accuracy is often low.

Therefore, EndoBRAIN, as an artificial intelligence system for assisting in the diagnosis of the endocytoscopy, has the advantage of rapid diagnosis. In the EC-NBI mode, it predicts as "Non-neoplastic" or "Neoplastic", and in the EC-stained mode, its prediction result is "Non-neoplastic", "Adenoma" or "Invasive cancer".

However, currently this artificial intelligence-assisted diagnostic system has not been applied in the Chinese population. The investigators plan to conduct a prospective clinical trial to validate the accuracy of EndoBRAIN for prediction of colorectal lesions histology in real-time endocytoscopy. This study will prospectively collect the lesions that meet the inclusion and exclusion criteria. After the endoscopic doctors make the diagnosis through endoscopic optics and EndoBRAIN, and then undergo endoscopic resection or surgical resection followed by pathological diagnosis, they will compare the doctor's diagnosis, the artificial intelligence diagnosis results with the gold standard pathological results, and summarize the diagnostic accuracy of this artificial intelligence-assisted diagnostic system for the colorectal lesions.

• Study Type: Observational
• Enrollment (Actual): 680

Contacts and Locations

Study Locations

• China
  • Jilin
    • Changchun, Jilin, China, 130021
      • The First Hospital of Jilin University

Participation Criteria

Eligibility Criteria

• Ages Eligible for Study: Child; Adult; Older Adult
• Accepts Healthy Volunteers: No

• Sampling Method: Non-Probability Sample

Study Population

The investigators analyzed only endoscopically or surgically resected colorectal lesions that had been observed with EC-NBI and EC-stained by endoscopists and artificial intelligence system before treatment that were ultimately performed histopathologic examination.

Description

Inclusion Criteria:

• Those patients who, during the endoscopic examination, discovered at least one colorectal lesion and received treatment and obtained a pathological diagnosis
• consent obtained for the study

Exclusion Criteria:

• non-epithelial tumors
• a history of inflammatory bowel disease
• chemotherapy or radiation therapy for colorectal cancer
• lesions without clear EC images
• specific pathological types
• familial adenomatous polyposis

Study Plan

How is the study designed?

Number of groups / cohorts

1

Cohorts and Interventions

Group / Cohort

Intervention / Treatment

Patients with one or more colorectal lesions detected; During endocytoscopy, the Clinician inspect for the presence of colorectal lesions as per routine clinical practice with the EndoBRAIN turned off. When a colorectal lesion is encountered, the Clinician will make a prediction on the histology based on routine clinical practice. Following this, the EndoBRAIN function will be switched on and the Clinician will take note of the EndoBRAIN prediction for the every image of colorectal lesion. In addition, other colorectal lesion features such as the size, location and shape will be recorded, which is similar to what is performed in routine clinical practice. The colorectal lesion will be resected and sent for pathological examination, which will form the "gold standard" for the diagnosis of colorectal lesion histology.

Diagnostic test: artificial intelligence system; The colorectal lesions had been observed with EC-NBI and EC-stained by endoscopists before treatment that were ultimately performed histopathologic examination. The endocytoscopies (CF-H290ECI, Olympus, Tokyo, Japan) have a maximum magnification of ×520, focusing depth, 35 μm; field of view, 570 × 500μm. During EC-NBI , the endoscopist pushed the button of the endoscope to switch from white-light imaging to NBI and observed the lesion with full magnification. After endocytoscopic observation, the artificial intelligence system will be open and display the predictive result. Finally, the endoscopist performed EC-stained mode diagnosis after staining the lesion surface with 1.0% methylene blue. After endocytoscopic observation, the artificial intelligence system will be open again and display the predictive result.; Other Names: endocytoscopy

What is the study measuring?

Primary Outcome Measures

Outcome Measure	Measure Description	Time Frame
To evaluate the diagnostic performance and high confidence diagnosis rate of EndoBRAIN in diagnosing neoplastic lesions in a clinical setting.EndoBRAIN in diagnosing neoplastic lesions in a clinical setting.	The sensitivity, specificity, accuracy, positive predictive value (PPV), negative predictive value (NPV), and high confidence diagnosis rate will be calculated for comparison with final histology as the gold standard for diagnosis	8 months

Secondary Outcome Measures

Outcome Measure	Measure Description	Time Frame
To evaluate the performance and high-confidence diagnostic rate of EndoBRAIN in diagnosing adenomas of rectosigmoid colon ≤5 mm;	The sensitivity, specificity, accuracy, positive predictive value (PPV), negative predictive value (NPV), and high confidence diagnosis rate will be calculated for comparison with final histology as the gold standard for diagnosis	8 months
To evaluate the performance and high-confidence diagnostic rate of Endobrain under EC-stained mode in the diagnosis of invasive cancer;	The sensitivity, specificity, accuracy, positive predictive value (PPV), negative predictive value (NPV), and high confidence diagnosis rate will be calculated for comparison with final histology as the gold standard for diagnosis	8 months
To evaluate the Influencing factors on the diagnosis of colorectal lesions by EndoBRAIN.	The diagnostic performance of EndoBRAIN in different Influencing factors will be calculated for comparison with final histology as the gold standard for diagnosis	8 months
To compare the diagnostic performance of diagnosing the histology of colorectal lesions by EndoBRAIN, by endoscopists, and by endoscopists combined with EndoBRAIN;	The diagnostic performance of diagnosing the histology of colorectal lesions by EndoBRAIN, by endoscopists, and by endoscopists combined with EndoBRAIN will be calculated for comparison with final histology as the gold standard for diagnosis	8 months

Collaborators and Investigators

• Sponsor: The First Hospital of Jilin University
• Investigators: Principal Investigator: Hong Xu, PHD, The First Hospital of Jilin University

Study record dates

Study Major Dates

• Study Start (Actual): April 1, 2024
• Primary Completion (Actual): December 19, 2024
• Study Completion (Actual): December 19, 2024

Study Registration Dates

• First Submitted: March 22, 2024
• First Submitted That Met QC Criteria: March 22, 2024
• First Posted (Actual): March 28, 2024

Study Record Updates

• Last Update Posted (Estimated): December 9, 2025
• Last Update Submitted That Met QC Criteria: December 2, 2025
• Last Verified: December 1, 2025

More Information

Terms related to this study

• Keywords: artificial intelligence; endocytoscopy
• Additional Relevant MeSH Terms: Neoplasms by Site; Neoplasms; Intestinal Diseases; Gastrointestinal Neoplasms; Digestive System Neoplasms; Digestive System Diseases; Gastrointestinal Diseases; Intestinal Neoplasms; Rectal Diseases; Colonic Diseases; Colorectal Neoplasms
• Other Study ID Numbers: 24K056-001

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