Clinical Decision Support for Ovarian Carcinoma Subtype Classification: A Pilot Observer Study With Pathology...

Area of Science:

• Digital Pathology
• Oncology
• Medical Informatics

Background:

• Clinical decision support (CDS) systems can aid pathologists with complex diagnostic tasks.
• The impact of CDS on pathologist performance for specialized tasks remains underexplored.

Purpose of the Study:

• To evaluate the effectiveness of a CDS tool in assisting pathology trainees with ovarian carcinoma subtype classification.
• To assess the impact of CDS on diagnostic accuracy and interobserver agreement using digital pathology.

Main Methods:

• Six pathology residents reviewed 90 whole slide images of ovarian carcinoma.
• Review was conducted both unaided and aided by a CDS tool identifying key histologic features.
• Performance was compared against a consensus reference standard from expert gynecologic pathologists.

Main Results:

• Aided review improved pairwise concordance with the reference standard for 5 of 6 observers (3.3%–17.8% increase).
• Mean interobserver agreement increased by 9.2% with CDS assistance.
• Observers benefited most from CDS prompts for previously missed, subtype-definitive histologic features.

Conclusions:

• Clinical decision support systems show potential for closing knowledge gaps in complex pathology diagnostics.
• CDS tools can enhance diagnostic accuracy and consistency for less experienced pathologists.