Visual and automated differential leukocyte counts. A comparison study of three instruments

Area of Science:

• Hematology
• Clinical Pathology
• Medical Technology

Background:

• Manual microscopic leukocyte counting is a standard diagnostic procedure.
• Automated methods offer potential for increased efficiency and standardization.

Purpose of the Study:

• To compare the accuracy and precision of two pattern-recognition leukocyte classifiers (Larc, Hematrak) and a cytochemical automated method (Hermalog D) against manual referee and technologist counts.
• To evaluate the impact of patient population and slide preparation on automated cell counting.

Main Methods:

• Comparison of cell counts from referee microscopy, multiple technologists, Larc, Hematrak, and Hermalog D.
• Testing with samples from random patients, selected abnormal patients, and healthy individuals.
• Analysis of factors contributing to count variability, including cell distribution on smears.

Main Results:

• All automated methods demonstrated good accuracy and effectively flagged abnormal results for review.
• Variability in pattern-recognition counts was primarily attributed to cell distribution on blood smears.
• Larc showed high sensitivity to cellular alterations, leading to slide rejections; Hemalog D exhibited superior precision.

Conclusions:

• Automated leukocyte counters are accurate and reliable for clinical use.
• Slide preparation techniques are critical for pattern-recognition instruments and influence instrument selection.
• Hermalog D offers the highest precision among the evaluated automated methods.