Dataset Overview
Definition：Bronchoscopy Endoscopy Video Imaging Datasetis provided byChangsha Langhui Information Technology Co., Ltd.The large-scale bronchial lenses built by Long Salanghui Information Technology Ltd. are constructed to check video image label databases. The complete recording of sound doors/pipes, trombones, left and right main bronchials and actual arrival of leaves, the retention of equipment to export original and complete inspection videos and key original maps, along with white light bronchor/ narrowband/self-flazing/ ultrasound bronchor multiple mode data, active checking/swipe/sap sections and full video association.
| Dataset Name | Bronchoscopy Endoscopy Video Imaging Dataset |
| Total Data Volume | Total Endoscopy Volume Included: 1,000,000 Cases |
| Imaging Modality | Endoscopy Video (White Light / NBI / Autofluorescence / EBUS) |
| Covered Body Parts | Sound door/airpipe/trunnel/right-and-right main bronchial/leave-part bronchial |
| Raw Data | Complete examination videos and key original images exported directly from the equipment |
| Main Protocols | White-light bronchial lenses; enhanced mode such as narrowband/self-fluorescent; ultrasound bronchos by special layer |
| Dedicated Fields | Arrival of bronchial level, stomatic area, mucous/clavic internal form, narrowness, haemorrhage, material extraction, pathology/cellology, microorganisms |
| Population Coverage | Oncology, infections, alien and post-operative population; mirrors/hard mirrors and EBUS layers |
| Source Institution | Cooperative Respiratory/Cultural Surgery at San Ace Hospital |
| Use Cases | Breather AI Navigation, Early Detection of Lung Cancer, EBUS Lead Plug-in AI, Categorization of Aeropathic Pathology |
Delivery Specifications and Field Requirements
The data set strictly follows the requirements of the ENDO-BRONCHO Unified Delivery Index to ensure data quality and traceability.
Delivery & Counting：One full bronchial mirror check per row; multi-series/multi-view only 1 case
Coverage：Record sound door/brain, tatter, left and right main bronchial and actual arrival of leaf; range, genre, haemorrhage and extraction of wood to be recorded
EBUS:Hypersonic bronchial lens EBUS by special layer; biopsy/swipe/sap segment association
Positives and Composition：Definite lesions / positive examinations as the main body；"Suspected", "considered", "cannot be excluded", "post-treatment changes" and "pathologically confirmed" are retained separately
Cross-Data Association：Must relate to official bronchoscopy reports; prioritize chest CT, pathology/cellulars, microorganisms and treatment
2026 Frontier AI Research Progress
Domain Review：BSAI made a breakthrough in the MASA Multi-According and Non-Suspecting Aberrations in 2025-2026, and EBUS AI and ROSE AI were close to expert level in sample sufficiency assessments, and research is evolving from labeling reliance to non-supervisory learning.
MSSA Multi-According SAM bronchial lens image partition
Frontiers in Oncology, 2026Integration of space, frequency and location information, simultaneous completion of the partition of the cookstove and diagnosis of lung cancer by a two-coder, validated using the BM-BronchoLC open data set.
KD-MFAD without supervision of bronchial lens tumour detection
2025The treatment of bronchial lens tumour recognition using knowledge distillation+ memory characteristics without supervised abnormality detection reduces reliance on a large amount of labeled data.
Summary of progress in AI intervention pulmonary medicine
2025EBUS AI and ROSE AI are close to expert level in sample sufficiency assessment, malignant cell testing and cellular sub-classification.
Bneumatic mirror AI system overview
2025AI has been used for bronchial lens image interpretation, pathological classification, real-time navigation and training assessment; EBUS AI has been used for lymphomy nodes and surrounding pathological profiling.
Annotation Workflow & Quality Control
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。
Initial Annotation (Specialist Physician)
Attending physicians annotate case by case against the standards, including lesion localization, morphological description and disease term determination。
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。
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
AI Frontier Research
BSAI made a breakthrough in the MASA Multi-According and Non-Suspecting Aberrations in 2025-2026, and EBUS AI and ROSE AI were close to expert level in sample sufficiency assessments, and research is evolving from labeling reliance to non-supervisory learning.