Mainecoon: Implementing an Open-Source Web Viewer for DICOM Whole Slide Images with AI-Integrated PACS for Digital...

Area of Science:

• Digital Pathology
• Artificial Intelligence in Medicine
• Medical Imaging Informatics

Background:

• Digital pathology faces challenges integrating diverse data formats and large AI annotations.
• Existing systems struggle with seamless AI integration into pathology imaging workflows.

Purpose of the Study:

• To develop an open-source solution (Mainecoon) for efficient AI integration in digital pathology.
• To address performance bottlenecks associated with AI-generated annotations on whole slide images (WSIs).

Main Methods:

• Developed Mainecoon, an open-source project utilizing the DICOM standard for WSIs.
• Integrated an AI model for Non-alcoholic steatohepatitis (NASH) detection, validated with the DICOM Workgroup 26 Connectathon dataset.
• Employed streaming and batch processing for enhanced data loading and frontend performance.

Main Results:

• Mainecoon successfully encodes AI results using the Microscopy Bulk Simple Annotations standard for seamless metadata integration.
• The system demonstrates improved data loading efficiency and reduced user waiting times.
• Web services implemented with Flask, integrated with a viewer and Raccoon archive, ensuring secure authentication.

Conclusions:

• The Mainecoon architecture offers a robust, interoperable, and practical solution for real-world digital pathology challenges.
• The project effectively addresses the integration of AI and structured data in WSIs.
• This approach enhances the performance and applicability of AI in pathology imaging workflows.