Exploring Data Cohorts in the REAL-Colon Dataset

The REAL-Colon dataset is revolutionizing the field of colonoscopy with its diverse and comprehensive collection of video recordings. This dataset combines endoscopy practices from various regions, providing a rich source of data for medical research and AI development.

The dataset includes recordings from clinical studies and acquisition campaigns conducted in different countries, ensuring a wide range of procedures and techniques are captured. Patients from these studies provided consent for their anonymized data to be used, and all necessary ethical approvals were obtained.

The video recording process utilized professional equipment to maintain high-quality footage, which was then compressed and converted into individual frames for analysis. Anonymization protocols were strictly followed to protect patient privacy and comply with GDPR and HIPAA regulations.

The dataset design involved a meticulous selection process to curate a set of 60 videos that represent a diverse range of colonoscopy procedures. Each video corresponds to an individual patient and may or may not contain instances of colon polyps.

Clinical data associated with each video includes patient demographics, technical details of the endoscope used, and the number of polyps resected during the procedure. The dataset also includes information on the cleanliness of the colon, evaluated using the Boston Bowel Preparation Scale.

Polyp detection annotation was carried out by a team of specialists under the supervision of expert gastroenterologists. The dataset includes detailed information on each polyp, including size, location, and histopathological results.

Overall, the REAL-Colon dataset provides a valuable resource for researchers and developers in the medical field, offering a wealth of information on real-world colonoscopy practices and polyp detection. This dataset has the potential to drive advancements in AI technology and improve patient outcomes in the field of gastroenterology.