Performance of the CellaVision advanced RBC application for schistocyte counting in the emergency clinical laboratory

Schistocyte quantification is essential for diagnosing microangiopathic hemolytic anemia (MAHA) and thrombotic microangiopathies (TMA). Early exclusion is critical for guiding appropriate management. However, the reference standard remains time-consuming manual microscopy, highly operator-dependent. In this study, the CellaVision Advanced RBC Application (ARBCA), an AI-based red blood cell morphology tool, was evaluated for schistocyte quantification in an emergency laboratory setting.

A total of 169 peripheral blood smears were analyzed in two tertiary-care laboratories. ARBCA pre-classification counts were compared with post-classification review by two expert laboratory specialists, who used them as the reference standard, following ICSH criteria. Agreement was assessed with Bland-Altman analysis, Passing-Bablok regression, and Lin's concordance correlation. Repeatability was evaluated on replicate smears, and diagnostic performance for MAHA was analyzed using ROC curves.

ARBCA showed overall agreement with expert results, with no significant systematic bias (mean difference 7.64 %; p=0.111). Passing-Bablok regression (slope 1.0; intercept 0.0) confirmed no proportional or constant error, except in MAHA cases, which deviated from linearity. Repeatability was acceptable (CV 8.0 %; IQR 2.80-12.40) and inter-rater agreement was high (κ=0.85). In MAHA patients, ARBCA underestimated schistocytes, misclassifying three cases. The optimal cutoff for MAHA prediction was ≥1.62 % (AUC 0.845; sensitivity 60.0 %, specificity 100.0 %), while expert review achieved an AUC of 1.00 with 100 % sensitivity and specificity.

ARBCA is a valuable screening tool that improves standardization, reproducibility, and turnaround time. However, expert morphological review remains essential, especially in emergencies where rapid and accurate decisions are lifesaving.