This paper reports on preliminary tests of an alternative computer technique for single step enumeration of both CD34-positive stem cells and marker beads detected within the 5-dimensional data space of the measured parameters: Forward Scatter, Side Scatter, CD45, CD34, 7AAD. We begin with ACluE, a general-purpose flow cytometry cluster analysis program without expert knowledge, and allow it to seek all cell clusters. The knowledge we add is that the clusters of interest are: 1. CD34+, CD45+, low SSC, medium FSC viable cells and 2. Marker beads.
The collections of data chosen for testing included single platform clinical results with a wide range of CD34 counts and sample viabilities and also control samples that are used routinely for validation.
The CD34+ clusters were shown to be reasonably distinct in the 5D data space and quite detectable by ACluE, even with positive counts as low as 16 in 146,000. In detailed examination, the very small proportion of “false positive” and “false negative” cells were deemed “borderline” by an experienced operator using manual analytical methods.
Marker beads are similarly readily detected and, in fact, distinct clusters identified as containing single and double beads were found; a factor that calculations should take account of.
The final question to be investigated is whether the conditions for cluster detection and merging may be fixed to allow analysis without operator intervention