Abstract
To evaluate performance of AI-assisted blast detection via whole blood film scanning in cases of severe leukopenia. From August to October 2023, we collected 157 patient specimens with white blood cell (WBC) counts of ≤2.0 × 10⁹/L. We then evaluated and compared the blast detection capabilities of an artificial intelligence-based system (Cygnus instrument) against the CellaVision DI-60 for WBC analysis. Additionally, this study aimed to assess the comparative performance of these two automated blood cell morphology analyzers based on their underlying operational principles. Following manual microscopy confirmation, 17 cases showing ≥ 1 blast cell in peripheral blood smear were identified among 157 specimens with WBC ≤ 2.0 × 10⁹/L. The Cygnus whole-slide scanning mode detected all 17 cases (sensitivity: 100%), whereas the Cygnus 200-cell mode detected only 9 cases (9/17), and the CellaVision DI-60 200-cell mode detected 8 cases (8/17). In leukopenic patients (WBC ≤2.0×10⁹/L), utilizing Cygnus' whole-slide scanning capability substantially improves blast identification rates, potentially enabling earlier detection of hematologic abnormalities.