Abstract
Digital morphology systems using artificial intelligence to pre-classify peripheral blood film white cell differentials are
increasingly deployed to reduce morphologist workload, but their practical value depends on how reliably the system's own
uncertainty estimate identifies the cells it should not classify without human review. We compared AI pre-classification against
consensus morphologist classification on 8,412 white cells from 210 blood films, and separately examined the system's
self-reported confidence score as a predictor of disagreement. Overall concordance with consensus morphology was 94.1%, but
concordance for cells the system flagged as low-confidence was only 71.3%, compared with 97.8% for high-confidence cells,
indicating that the confidence score usefully separates reliable from unreliable automated calls. We report the cell types most
often driving low-confidence flags and discuss what this pattern implies for how a laboratory should configure the review
threshold rather than relying on overall concordance alone.
References
Rodellar J, Barrera K, Alferez S, et al. "A deep learning based system for automatic classification of peripheral blood cell images". Comput Biol Med 143(2022): 105233.
Vandenberghe ME, Scott MLJ, Scorer PW, et al. "Relevance of deep learning to facilitate the diagnosis of HER2 status in breast cancer". Sci Rep 7(2017): 45938.
Barnes A, Mangion J, Aslan D, et al. "Development of a digital hemocytology system: a proof-of-concept study". Int J Lab Hematol 41(2019): 685-693.
Kratz A, Bengtsson HI, Casey JE, et al. "Performance evaluation of the CellaVision DM96 system: WBC differentials by automated digital image analysis supported by an artificial neural network". Am J Clin Pathol 124(2005): 770-781.
Briggs C, Longair I, Slavik M, et al. "Can automated blood film analysis replace the manual differential? An evaluation of the CellaVision DM96 automated image analysis system". Int J Lab Hematol 31(2009): 48-60.
Merino A, Puigvi L, Boldu L, et al. "Optimizing morphology through blood cell image analysis". Int J Lab Hematol 40(2018): 54-61.
Bercovitz RS, Lezotte D, Quinones RR, et al. "Comparison of manual and automated differential leukocyte counts: a systematic review". Int J Lab Hematol 34(2012): 617-626.
Rumke CL. "Imprecision of ratio-derived differential leukocyte counts". Blood Cells 11(1985): 311-314.
Faucher JL, Lacronique-Gazaille C, Frebet E, et al. "6 markers/5 colors extended white blood cell differential by flow cytometry". Cytometry A 71(2007): 934-944.
Ceelie H, Dinkelaar RB, van Gelder W. "Examination of peripheral blood films using automated microscopy; evaluation of Diffmaster Octavia and Cellavision DM96". J Clin Pathol 60(2007): 72-79.
This work is licensed under a Creative Commons Attribution 4.0 International License.