Polyp-Size: A Precise Endoscopic Dataset for AI-Driven Polyp Sizing.

Area of Science:

• Gastroenterology and Medical Imaging
• Artificial Intelligence in Medicine

Background:

• Accurate colorectal polyp size is critical for clinical management but challenging due to subjective measurement methods.
• Existing artificial intelligence (AI) datasets for polyps often lack precise size annotations, hindering the development of robust size estimation models.

Purpose of the Study:

• To introduce Polyp-Size, a novel dataset of colonoscopy videos with validated polyp size measurements.
• To facilitate the development of AI models for accurate automated polyp size estimation.

Main Methods:

• Collected 42 high-resolution white-light colonoscopy videos.
• Measured polyp sizes post-resection using vernier calipers to submillimeter precision.
• Annotated polyps with validated size, Paris classification, anatomical location, and histological type.

Main Results:

• The Polyp-Size dataset offers submillimeter precision in polyp size measurements.
• Includes diverse polyp characteristics and standardized video formats.
• Enables the training of AI models specifically for polyp size estimation.

Conclusions:

• The publicly available Polyp-Size dataset addresses a critical gap in AI-driven medical imaging for colorectal cancer screening.
• This resource will foster research and collaboration to improve automated polyp measurement accuracy in clinical practice.