Neural network analysis of digital flow cytometric data

In flow cytometry, the pulse waveform features measurable by current analog instruments are limited to the pulse integral, peak, and width. Digitalization of the waveforms provides a means for the extraction of additional features, such as skewness, kurtosis, and Fourier properties. The introduction of additional features requires automated procedures for classification of biological particles. In this work, we implemented and evaluated neural network classification algorithms using derived, complex features, as well as using the raw, sampled data without feature extraction. The performance of the neural networks was compared with that of a more conventional means of classification in flow cytometry, the K-means clustering algorithm.