Artificial intelligence-based triage of large bowel biopsies can improve workflow

Area of Science:

• Digital pathology
• Artificial intelligence in healthcare
• Gastrointestinal pathology

Background:

• Large bowel biopsies are common diagnostic specimens.
• A service evaluation was conducted to assess AI feasibility for triaging these biopsies.
• The goal was to prioritize urgent cases from a reporting backlog.

Purpose of the Study:

• To evaluate the feasibility of an AI system for triaging large bowel biopsies.
• To improve reporting turnaround times for critical findings.
• To integrate AI into routine pathology workflows.

Main Methods:

• An AI platform was developed and interfaced with slide scanner and reporting software.
• Pathologists provided feedback to refine the AI model.
• The AI classified biopsies into diagnostic categories, prioritizing neoplasia and inflammation.

Main Results:

• The AI achieved high sensitivity and specificity for diagnosing neoplasia and significant pathology.
• The AI triaged pathway significantly reduced reporting turnaround times for neoplastic and inflammatory cases.
• The automated pathway processed each slide in approximately 175 seconds.

Conclusions:

• National Health Service (NHS) laboratory staff successfully implemented an AI solution for large bowel biopsy triage.
• The AI system accurately classified biopsies, improving reporting efficiency for key diagnoses.
• The integration enhanced turnaround times for urgent cases, demonstrating AI's value in pathology.