Performance Evaluation of a Novel Artificial Intelligence-Assisted Digital Microscopy System for the Routine Analysis...

Area of Science:

• Hematology
• Computational Pathology
• Medical Diagnostics

Background:

• Bone marrow aspiration (BMA) smear analysis is crucial for diagnosing and monitoring hematological conditions.
• Current BMA analysis relies on manual microscopy, which can be time-consuming and subjective.
• There is a need for more efficient and objective methods for BMA evaluation.

Purpose of the Study:

• To validate a computational microscopy approach with an artificial intelligence (AI)-driven decision support system for BMA analysis.
• To compare the performance of the AI system against manual microscopy for BMA smear evaluation.

Main Methods:

• A multicenter study analyzed 795 BMA specimens (Romanowsky-stained and Prussian blue-stained).
• The Scopio Labs X100 Full Field BMA system (test method) was compared with manual microscopy (reference method).
• Efficiency, sensitivity, specificity, and interuser agreement were evaluated for various BMA characteristics.

Main Results:

• The AI system demonstrated high correlation with manual microscopy for comprehensive BMA evaluation.
• Efficiency was 90.85% for Romanowsky-stained and 90.0% for Prussian blue-stained samples.
• Overall agreement between the test and reference methods was 91.1%, with excellent repeatability and reproducibility.

Conclusions:

• The AI-driven digital decision support system provides high-quality, accurate digital BMA analysis.
• This system has the potential to expedite expert review and diagnosis of BMA specimens.
• Practical applications include remote BMA evaluation and advancing research in hematopoiesis.