AI Research Analysis
Curated Endoscopic Retrograde Cholangiopancreatography Images Dataset
Endoscopic Retrograde Cholangiopancreatography (ERCP) is a key procedure in the diagnosis and treatment of biliary and pancreatic diseases. While Artificial Intelligence is a promising solution for automation, public ERCP datasets are scarce, hindering progress. This study addresses this gap by providing a large, curated dataset to advance AI in this critical medical field.
Key Metrics & Impact
This study provides a large, curated ERCP image dataset to fill a critical gap for AI development in pancreaticobiliary endoscopy. It includes 19,018 raw and 19,317 processed images from 1,602 patients, with 5,519 manually labeled by experienced gastroenterologists. The dataset's utility is validated through classification experiments, achieving 78.3% accuracy with EfficientNet-B7, establishing a strong benchmark for automated ERCP analysis.
0
Raw Images Collected
0
Processed Images
0
Patients Represented
0
Expert-Labeled Images
Deep Analysis & Enterprise Applications
Select a topic to dive deeper, then explore the specific findings from the research, rebuilt as interactive, enterprise-focused modules.
Automated Image Processing Workflow
DICOM Extraction
→
Manual S/V Classification
→
Canny Edge Detection
→
Vertical Line Detection
→
Image Partitioning
→
Quality Validation
→
Anonymization & Labelling
Diverse Image Acquisition: The dataset includes images from various fluoroscopic systems (Ziehm Vision RFD 3D, Philips PCR Eleva), ensuring natural variability and robustness for AI models. It covers all ERCP workflow stages from pre-cannulation to post-procedural documentation.
5,519
Expert-Labeled Images
A substantial subset of images was manually annotated by two gastroenterologists (5+ years experience) and reviewed by a senior clinician (20+ years experience), ensuring high diagnostic reliability.
Model Performance Benchmarking: The dataset's utility was validated through a multi-model classification experiment to distinguish between lithiasis, bile leaks, strictures, and normal findings.
| Architecture | Accuracy | Precision (macro) | Recall (macro) | F1-score (macro) | F1-score (validation) |
|---|---|---|---|---|---|
| MobileNetV2 | 61.4 | 58.7 | 60.5 | 0.574 | 0.550 |
| EfficientNet-B7 | 78.3 | 75.3 | 75.3 | 0.738 | 0.609 |
| ResNet50 | 63.7 | 61.7 | 62.1 | 0.617 | 0.490 |
| DenseNet121 | 65.9 | 66.0 | 63.1 | 0.621 | 0.590 |
| DeiT3-Small | 46.0 | 11.5 | 25.0 | 0.158 | 0.148 |
Robust Splitting & Anonymization: Images were split into training, validation, and test sets using a stratified strategy based on individual ERCP exams to prevent data leakage. Patient identifiers were anonymized via central square crop and circular masks, and original IDs replaced with anonymized indices.
Empowering Next-Gen Pancreatobiliary AI
Problem: Despite growing interest in AI for pancreaticobiliary endoscopy, development is hindered by a scarcity of publicly available, high-quality fluoroscopic ERCP datasets. This limits reproducibility, generalization, and benchmark creation.
Solution: This dataset directly addresses the data limitation by providing a large, curated, multi-system collection of ERCP fluoroscopic images, including diagnostic labels. It enables standardized benchmarks, accelerates research in image classification, lesion detection, and procedural quality assessment.
Impact: Facilitates the development of more robust and generalizable AI models for crucial diagnostic and therapeutic tasks in ERCP, ultimately improving patient outcomes and procedural efficiency.
Ethical Approval & Anonymization: The study was approved by the Board of Directors of Unidade Local de Saúde do Alto Minho (ULSAM), Approval ID: 243/CA-2025. All patient identifiers were meticulously removed or anonymized to ensure privacy and compliance with ethical guidelines.
Data Availability & Reproducibility: The dataset is publicly available on Figshare, and all preprocessing code is available on GitHub. This promotes transparency, reproducibility, and allows researchers to adapt or extend the processing workflow for their own data and experimental settings.
Calculate Your Potential AI ROI
Estimate the significant time and cost savings your enterprise could achieve by implementing AI solutions tailored to your industry.
Estimated Annual Savings
Annual Hours Reclaimed
Your AI Implementation Roadmap
A structured approach to integrating AI into your enterprise, ensuring maximum impact and seamless adoption.
Phase 01: Discovery & Strategy
Comprehensive assessment of current operations, identification of AI opportunities, and development of a tailored strategy aligned with your business objectives.
Phase 02: Data Preparation & Model Training
Collection, cleaning, and preparation of relevant data. Selection and training of appropriate AI models, including custom architecture development if required.
Phase 03: Integration & Deployment
Seamless integration of AI models into existing systems and workflows. Robust deployment with rigorous testing and validation to ensure performance and reliability.
Phase 04: Monitoring & Optimization
Continuous monitoring of AI model performance, ongoing optimization, and iterative improvements to adapt to evolving business needs and data patterns.
Ready to Transform Your Enterprise with AI?
Book a free consultation with our AI specialists to discuss how these insights can be leveraged to drive innovation and efficiency in your organization.
Ready to Get Started?
Book Your Free Consultation.
Let's Discuss Your AI Strategy!
Lets Discuss Your Needs
Select a Date
Sun
Mon
Tue
Wed
Thu
Fri
Sat