A knowledge-based system for the interpretation of flow cytometry data in leukemias and lymphomas

Area of Science:

• Hematopathology
• Computational Pathology
• Clinical Diagnostics

Background:

• Flow cytometry immunophenotyping and DNA analysis are crucial for diagnosing hematopoietic disorders.
• A shortage of physicians trained in interpreting these complex flow cytometry studies exists.
• AI-powered tools can assist in standardizing and improving diagnostic accuracy.

Purpose of the Study:

• To evaluate the performance of a knowledge-based computer system, Professor Fidelio, in interpreting flow cytometry and DNA analysis results for hematopoietic diseases.
• To assess the system's utility as a stand-alone diagnostic aid.

Main Methods:

• Professor Fidelio, a heuristic classification system, was tested on 366 patient specimens.
• The system's interpretations were compared against the established diagnoses from two tertiary medical centers.
• Discrepancies were analyzed to identify potential sources of error or differences in diagnostic criteria.

Main Results:

• Professor Fidelio achieved an 82% agreement rate with the diagnosis of record across all tested specimens.
• Interpretations were deemed appropriate in all cases, with most disagreements attributed to diagnostic recording errors or minor criterion variations.
• The system demonstrated limitations in specificity for certain lymphoproliferative disorders requiring morphologic data.

Conclusions:

• Professor Fidelio is a valuable tool for aiding in the interpretation of flow cytometry and DNA analysis in hematopathology, particularly in settings with limited expert availability.
• The system's integration into a broader diagnostic workstation enhances its utility for reporting, education, and research.
• Further refinement may be needed for complex subclassifications that rely heavily on morphology.