Development and operation of a digital platform for sharing pathology image data.

Area of Science:

• Digital pathology
• Artificial intelligence in medicine
• Medical data sharing

Background:

• Artificial intelligence (AI) research heavily relies on data availability.
• The medical field's increasing adoption of AI drives demand for high-quality medical data.
• Development of a digital pathology data platform for AI researchers is described.

Purpose of the Study:

• To develop and describe a platform for providing and sharing digital pathology data.
• To highlight challenges in operating a sustainable data-sharing platform with pathologists.
• To meet the increasing demand for quality medical data in AI research.

Main Methods:

• A dataset of over 3000 pathological slides from five organs in tumor cases was curated.
• Pathologists annotated tumor areas on digitized slides to create ground truth for AI training.
• AI-assisted annotation was implemented to reduce pathologist workload, in collaboration with AI teams.

Main Results:

• A web-based data sharing platform was launched in 2019, featuring 3100 images.
• The platform provides 5 pre-processing algorithms for AI researchers.
• The platform facilitates the sharing of massive pathological image data.

Conclusions:

• The medical image sharing platform effectively meets AI developers' demand for quality data, despite development challenges.
• Patient consent during data acquisition is crucial for international data sharing platforms due to privacy regulations.
• The study aims to guide future researchers in creating more effective and accessible medical data platforms.