Dataset Overview

Definition:Colonoscopy Endoscopy Video Imaging Datasetis provided byChangsha Langhui Information Technology Co., Ltd.The large-scale colonoscopy inspection video image representation database, built by Long Salang Langhui Information Technology, Ltd., is built. Full records are entered, reaching the deepest part, back-to-back the blind/end retrospect mark, back-scans and the colon sections, the equipment is retained to export original full-screen video and key original maps, and the corresponding intestinal intestinal scoring, stagger size/form/spectrum, removal of integrity and pathology validation data.

Dataset NameColonoscopy Endoscopy Video Imaging Dataset
Total Data VolumeTotal Endoscopy Volume Included: 1,000,000 Cases
Imaging ModalityEndoscopy Video (White Light / NBI / Magnification / Chromoendoscopy)
Covered Body Partsrectal/betatis/remote colon/wire colon/litre/remote blind/remote
Raw DataComplete examination videos and key original images exported directly from the equipment
Dedicated FieldsIntestinal preparation scoring, arrival, retrospection time, salivation size/form/spectation, removal integrity, pathology, complications
Main ProtocolsWhite-ray endoscopy; narrowband/optic enhancement; magnification/dye endoscopy; celibate/livespect
Pathology CorrelationMust relate to official endoscopy reports; prioritize pathology, CT/MRI, treatment and follow-up
Source InstitutionGastroenterology Department / Endoscopy Center of Partner Grade-A Tertiary Hospitals
Use Casesintestinal mirrors Auxiliary diagnosis, real-time detection and division of salivated meat, adenoma detection rate, CRC screening

Delivery Specifications and Field Requirements

The data set strictly follows the requirements of the ENDO-COLONO Unified Delivery Index to ensure data quality and traceability.

Delivery & Counting:One full colonoscopy in one row; multi-series/multi-view only 1 case

Coverage:Record entry, reach the deepest part, return to the blind/end of the colon mark (if arrived), exit mirror and colon section

Intestine preparation:Intestine preparation and retrospect integrity must be recorded; standardized ratings such as BBPS are used

Positives and Composition:Definite lesions / positive examinations as the main body;Normal complete examinations, postoperative follow-up examinations and quality-limited examinations are grouped separately

Cross-Data Association:Multiple perspectives of the same size cannot be counted; cross-modular linkages are only patient-level or examination-level connections without pixel-level characterization

2026 Frontier AI Research Progress

Domain Review:The intestinal mirror AI is the most mature area of clinical deployment in medical endoscope AI: Olympus Cadie has been confirmed by FDA-clared and CE-marked, EAGLE tests to increase the absolute value of ADR by 7.3 per cent; PESNet achieves a real-time partition of 12.6 ms/frame.

EAGLE experiment: Olympus Cadie AI

npj Digital Medicine, 2025

The CDTs in the multi-centres include 841 patients. The caddie glandoma detection rate (ADR) has increased by 7.3% in absolute terms; the cadenoma detection rate has increased by 93%, the non-salent cardiac adenoma by 57% and the SSL by 230%.

PESNet real-time stasis and precision.

Frontiers in Oncology, 2025

End-to-end delay of 1080 p to 12.6 ms/frame to meet the clinical budget of 40 ms; reduction of quenching by 26 per cent and reduction of the edge of residual tumors after cold cutting by 15 per cent.

StileGAN Enhanced Spectrum Testing Broaden

npj Digital Medicine, 2025

Using StyleGAN to synthesize over 150,000 new bioimages of the colon, YOLOv5 detection mAP was raised from 0.86 to 0.93 to reduce the gap in external validation across the board.

AIS-Assisted enteroscope centre RCT

Gut and Liver, 2025

The difference is significant, both in the rate of 72.2% vs. 54.5% for the cavity of the CADe and in the rate of adenomas at 52.3% vs. 36.1%.

Annotation Workflow & Quality Control

1

Image Acquisition & De-identification

Standardized acquisition workflow: complete data is exported directly from the equipment, and patient identifying information (PHI) is removed before the data is stored。

2

Initial Annotation (Specialist Physician)

Attending physicians annotate case by case against the standards, including lesion localization, morphological description and disease term determination。

3

Review (Associate Chief Physician or Above)

Experts with the title of Associate Chief Physician or above review the initial annotation results item by item, correcting erroneous annotations and supplementing missing dimensions。

4

Consistency Assessment

10% of samples are randomly selected and independently annotated by 3 physicians to calculate Fleiss' Kappa; below 0.Dimensions scored 75 are flagged for rework。

License and Usage Agreement

Academic Research License

For universities and research institutions, supporting academic research related to medical AI。After signing the agreement, a de-identified data subset is provided; the source must be acknowledged:Langhui Technology DataAssetsAPI。

Commercial License

For medical device companies and AI diagnostics companies, supporting medical AI product R&D and medical device registration。Provides full datasets + custom annotation + incremental update services。

Data Compliance Statement

All images come from legally authorized sources; PHI fields are fully de-identified and contain no information that can directly identify an individual, in compliance with the Personal Information Protection Law and the Data Security Law。

Customized services

Supports extended requirements such as adding specific disease types, multimodal annotation expansion, and custom training of AI-assisted diagnostic models。

Quick Facts

Dataset CodeENDO-COLONO
Total Data Volume1 Million Cases
Imaging ModalityEndoscopy Video (White Light / NBI / Magnification / Chromoendoscopy)
Source InstitutionPartner Grade-A Tertiary Hospitals
Quality Control StandardsKappa ≥ 0.75
Compliance & LicensingEnd-to-End Compliance

AI Frontier Research

The intestinal mirror AI is the most mature area of clinical deployment in medical endoscope AI: Olympus Cadie has been confirmed by FDA-clared and CE-marked, EAGLE tests to increase the absolute value of ADR by 7.3 per cent; PESNet achieves a real-time partition of 12.6 ms/frame.

Latest 2025–2026 Papers
Nature / Nature Medicine-level Research
FDA / NMPA / CE Regulatory Updates