Computer-aided simple triage.

Area of Science:

• Medical Imaging
• Artificial Intelligence in Healthcare
• Radiology Workflow Optimization

Background:

• Computer-aided detection (CAD) traditionally serves as a second reader in medical imaging to enhance diagnostic accuracy.
• Advancements in CAD technology and expanded imaging modality coverage necessitate exploring new roles for CAD systems.

Purpose of the Study:

• To investigate the paradigm shift of CAD systems in emergency diagnostic imaging from a second reader to an initial interpreter and triage tool.
• To define and explore Computer-aided Simple Triage (CAST) systems for emergency diagnostics.

Main Methods:

• Extracted common characteristics of existing CAD systems and compared them to emergency diagnostic imaging modalities.
• Defined a new class of CAD systems, CAST, based on deduced requirements for emergency triage.
• Explored CAST properties, use case scenarios, clinical benefits, and differentiated it from CAD and automated computer diagnosis.

Main Results:

• A CAST system should provide fully automatic, per-study classification for simple triage.
• CAST enables immediate expert review for positive studies, reducing critical patient delays.
• Negative studies can be initially managed by less experienced staff, optimizing resource allocation.
• Automatic assessment of image quality and study complexity can guide reading prioritization.
• CAST systems require high specificity without compromising high per-study sensitivity.

Conclusions:

• CAST systems can act as enabling technology for integrating advanced imaging into emergency protocols.
• CAST addresses challenges of reader unavailability and reading prioritization in emergency radiology.
• This approach has the potential to significantly improve emergency diagnostic imaging workflows.