Poster
BCM
Artificial Intelligence at the Point of Care: Comparative Performance of Two Digital Microscopy Analysers for Acute Leukaemia Detection
miLab™ showed better acute leukemia alerting than HemoScreen™ and performed well for routine blood counts in a multicenter ED sample study.
PoCT Symposium
|March 26, 2026
|Manon Alexandre, et al.
Abstract
This multicentre study compared AI-based POCT analysers miLab™ BCM and HemoScreen™ against Sysmex XN9000 for routine CBC and acute leukaemia (AL) detection in ED settings, with miLab showing superior blast flagging (100% sensitivity for circulating blasts).
| Parameters | WBC | PLT | Neutrophil | Blast | |
|---|---|---|---|---|---|
| Category | Pearson Correlation Coefficient | Flag Sensitivity | Flag Specificity | ||
| miLab™ BCM | >0.92 | >0.92 | >0.92 | 100% | 92% |
| HemoScreen™ | >0.92 | >0.92 | >0.92 | 20% | 98% |
Table 1: Pearson Correlation and Flag Performance of miLab™ BCM vs. HemoScreen™
Key Highlights
- Multicenter study of 60 EDTA samples, including 22 hematologic malignancies and 7 acute cases.
- Compared updated HemoScreen™ and miLab™ BCM against the Sysmex XN9000 reference analyzer.
- Both devices correlated well for WBC, platelet, and neutrophil counts.
- miLab™ "Blast" flag: 71% sensitivity and 92% specificity; HemoScreen™ "ABN" flag: 14% sensitivity and 98% specificity.
- When cases without circulating blasts were excluded, miLab™ reached 100% sensitivity, while HemoScreen™ remained at 20%.
- miLab™ also allowed direct or remote access to digital images for expert review.
Keywords
acute leukaemia
POCT
blast flag
AI digital microscopy
CBC