What happened
CellaVision of Sweden makes the Advanced RBC Application, an optional module for its DM96 and DM1200 analyzers. It uses trained artificial neural networks to pre-characterize red cells by size, color, shape and inclusions into 21 morphological characteristics. Operators confirm or change each suggestion. Method comparison studies ran at five clinical sites, two in the United States and three outside it. The main comparison with manual microscopy covered 1181 samples. FDA reviewers report that the results met the predefined acceptance criteria. For schistocytes, target cells, sickle cells, tear drop cells and parasites, efficiency was 90.6%, sensitivity 85% and specificity 93.6%. Agreement on cell size with an automated counter was lower, at 81.3%. FDA cleared the module on 1 August 2017. The results include operator review, so the accuracy of the networks alone is not reported.
Sources
-
Regulatory document, by authors with no tie to the system.
To determine agreement between the Advanced RBC Application and manual microscopy reference method, the efficiency, analytical sensitivity and analytical specificity were calculated along with the 95% confidence intervals for color, size and shape. Comparison between the DM system with Advanced RBC Application and the automated differential cell counter method for size was provided as overall agreement, positive percent agreement (PPA) and negative percent agreement (NPA) along with the 95% confidence interval. The results met the predefined acceptance criteria.
Details
AI system
- Workflow
-
Venous blood was collected in K3EDTA tubes, and one slide was prepared from each sample. At one site an expert operator confirmed or overrode the suggested grading, and at the others operators re-characterized individual cell images.
Follow-up
- Conditions for replication
-
Skilled operators trained on the device and in recognizing blood cells review the suggestions. The blood samples have been flagged as abnormal by an automated cell counter.
Entry details
- Sector
-
Healthcare.
- Purpose
-
CellaVision built the Advanced RBC Application to pre-characterize red blood cell morphology on blood smears for skilled operators to confirm or modify.
Not yet checked: model, release, provider, role, configuration, inputs and outputs, autonomy, human oversight, AI contribution, beneficiaries, reach, distribution and access, duration, every measurement question, every money question, ownership and control, dependencies, cross-border connections, geopolitical implications, replication attempts, availability, institutional decision, grounds, action and result, success criterion, every evidence question, observed or intended, alternative assessments, related entries, other identifiers, category mappings.
Cite this record
Peter H. Vartanian. (2026). Advanced RBC Application sorted red cells close to manual microscopy (AIBD-0805). Artificial Intelligence Benefits Database, edition of 31 October 2026. https://petervartanian.xyz/aibd/records/AIBD-0805.html (archived at https://doi.org/10.7910/DVN/AADSII, version 4.0)