NDER: A novel web application using annotated whole slide images for rapid improvements in human pattern recognition

Area of Science:

• Digital Pathology
• Medical Education Technology

Background:

• Whole-slide images (WSIs) offer valuable data for medical training but are often underutilized.
• Current methods do not fully leverage WSI advantages for novel applications.
• A new pipeline was developed to convert annotated WSIs into a training and quality assurance tool.

Purpose of the Study:

• To establish an efficient workflow for extracting annotated WSIs.
• To integrate these images into the Novel Diagnostic Electronic Resource (NDER) web application.
• To facilitate high-throughput pattern recognition training.

Main Methods:

• Two image extraction and database creation methods were compared.
• Method 1 involved manual extraction and multi-pathologist validation.
• Method 2 used single-pathologist validation of annotated regions and automated image extraction.
• Annotated WSIs were classified into five diagnostic categories.

Main Results:

• The NDER workflow is highly automated and efficient, requiring minimal human input.
• Annotation takes 5 minutes, and expert validation takes 1-2 minutes per WSI.
• NDER effectively presents hundreds of high-resolution images with immediate user feedback.
• The system supports rapid pattern recognition and diagnostic proficiency evaluation.

Conclusions:

• The developed pipeline and NDER application provide an efficient method for utilizing annotated WSIs.
• This approach significantly enhances pattern recognition training and diagnostic skill assessment.
• NDER offers a valuable tool for medical education and quality assurance in pathology.