CELLAVISION DM96 AUTOMATIC HEMATOLOGY ANALYZER

The CellaVision DM96 is an automated cell-locating device that utilizes slide-scanning microscopy and image analysis software, including artificial neural networks, to locate and pre-classify cells on peripheral blood smears. It is designed to aid skilled operators in performing white blood cell differential counts, red blood cell morphology characterization, and platelet estimation, with final cell identification and classification requiring human verification.

The CellaVision DM96 is an automated hematology analyzer designed to assist in the analysis of peripheral blood smears. The system comprises a slide-scanning unit, which includes a slide feeder, microscope, and camera, coupled with a computer system running 'CellaVision Blood Differential software'. Its primary functions are to automatically locate white blood cells, characterize red blood cell morphology, and facilitate platelet estimation. The device captures digital images of blood cells from stained slides, and the software then suggests a classification for each white blood cell, typically using methods including deterministic artificial neural networks. The DM96 is intended for use by skilled operators who are trained in both the use of the device and in the recognition of blood cells. A crucial aspect of its methodology is that the operator must review, identify, and verify the software's suggested classification for each cell. For red blood cells and platelets, the system presents an overview image from which the operator characterizes morphology and estimates platelet concentration. Clinical evaluations have demonstrated its equivalence to predicate methods and devices in terms of accuracy for cell-location, verified classification accuracy and precision for each cell class, clinical sensitivity and specificity, and overview image quality.

AI Regulatory Concerns

Powered by

Jev reads the full FDA letter and returns calibrated concern probabilities per regulatory category. These are AI estimates, not regulatory advice.

• Q1. What specific algorithms or features within the 'CellaVision Blood Differential software' contribute to the pre-classification of white blood cells, and what is its standalone accuracy before operator verification?; How does the DM96's performance in cell location and classification vary with different types of Romanowsky stains or variations in smear preparation quality?; What is the typical throughput of the DM96 system (e.g., slides per hour) when including the operator verification step, and how does this compare to fully manual methods?; Are there any known limitations of the DM96 in identifying or classifying specific rare or highly abnormal cell morphologies?; What were the quantitative metrics used to define and confirm 'overview image quality' for red blood cell morphology and platelet estimation during clinical tests?