We have developed a method to assess the overall quality of a cell counting measurement. This method does not require a reference material or "ground truth" cell number value and is independent of cell type and measurement technique/instrument.
To quantify the performance metric for a counting measurement process, we assess two main properties to determine the overall quality of the measurement: proportionality and precision. A measurement adheres to proportionality when a change in the input parameter (in this case, dilution fraction) is always accompanied by a change in the measurand (in this case cell number), and that change is always related by a constant multiplier. In cell counting, this fundamental principle implies that the measured cell number should be proportional to the dilution fraction under ideal experimental conditions. This general concept of proportionality must hold true for a measurement process to be accurate (i.e. behave without bias from the measurement process); deviation from proportionality would likewise indicate that a bias (systematic or non-systematic) has occurred to reduce the overall measurement confidence. The performance metric provides a relative measure of accuracy for cell counting by assessing closeness of agreement between the obtained cell count data and the fundamental property of proportionality. Measurement precision also provides another measure of the overall quality of the measurement process. Highly scattered experimental data with average cell numbers fitting well to proportionality would similarly reduce the overall robustness of the measurement process.
Through appropriate experimental design and statistical analysis, a performance metric can be calculated based on the deviation of cell count values (i.e. residuals) from a modeled ideal proportional response. The performance metric is sensitive to both closeness to proportionality as well as random variations occurring from the observation process (i.e. precision). Additional metrics (i.e. bias, linearity, percent recovery etc.) can also be determined from the experimental design and may be useful in cell-count process validation.
The performance metric derived from the experimental and statistical framework characterizes the entire cell count measurement process, including the measurement platform, method specific factors such as dilution steps and sampling, and the specific cell preparation measured.
The experimental design and statistical analysis, developed in close collaboration with the NIST Statistical Engineering Division, is under consideration as an International Standard through the International Standards Organization (ISO)/Technical Committee (TC) 276 – Biotechnology. A general guide for cell counting is also being developed within ISO/TC 276. Details of this work have been showcased at a TERMIS pre-conference workshop highlighting public/private collaboration, disseminated through an ISCT webinar and several other conferences and workshops.
Contact Sumona Sarkar at sumona.sarkar [at] nist.gov () or Steven Lund at steven.lund [at] nist.gov () to learn more about this project.