Reproduce “Curated endoscopic retrograde cholangiopancreatography images dataset” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Alda João Andrade, Mónica Martins, André Ferreira, Tarcísio Araújo, Luís Lopes, Victor Alves
Researchers developed a large, expertly curated, and publicly available dataset of Endoscopic Retrograde Cholangiopancreatography (ERCP) fluoroscopic images derived from routine clinical practice. A technical validation demonstrated the dataset's utility by achieving an F1-score of 0.738 with an EfficientNet-B7 model for diagnostic classification.
Key insights
- A systematic methodology combining retrospective data extraction from a high-volume clinical center with a robust multi-stage processing pipeline can generate a large, diverse, and clinically relevant medical imaging dataset.
- Expert manual annotation, cross-checked against procedural reports and reviewed by senior clinicians, is essential for creating a high-quality labeled dataset suitable for supervised AI tasks in complex medical domains.
- Rigorous anonymization protocols are crucial for enabling public sharing of sensitive medical imaging data, thereby fostering open science and collaborative AI research.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.