RT Journal Article
SR Electronic
T1 A Seven-Parameter Approach to White Blood Cell Identification
JF American Society for Clinical Laboratory Science
JO Clin Lab Sci
FD American Society of Chemistry and Laboratory Science
DO 10.29074/ascls.2023003225
A1 Memmott, Audrey
A1 Tanner, Dylan
A1 Cordner, Ryan
YR 2024
UL http://hwmaint.clsjournal.ascls.org/content/early/2024/11/26/ascls.2023003225.abstract
AB Medical laboratory scientists routinely evaluate blood smears in the clinical laboratory. A high level of skill in white blood cell morphology is required to do the task quickly and accurately. Despite significant training and skill, medical laboratory scientists still demonstrate variability and some inaccuracies in their cell morphology skills. In this study, we proposed a series of 7 questions that can guide morphological identification of white blood cells. We validated the potential of this approach with a random forest classifier algorithm using a dataset of 3600 cells. With the 7 categories as input, the random forest model was able to classify white blood cells with an overall accuracy of 98.7%. We further demonstrated that this approach has application for medical laboratory scientists with color blindness by testing a color independent model of white blood cell identification using 5 parameters. Our approach has the potential to improve educational approaches to white blood cell morphology and could increase the consistency of manual differential results between medical laboratory scientists.