Application of neural networks to flow cytometry data analysis and real-time cell classification

Area of Science:

• Biotechnology
• Computational Biology
• Data Science

Background:

• Traditional flow cytometry analysis relies on graphical, two-parameter scatter plots, which become cumbersome with increasing data dimensions.
• The geometric increase in plots with more measured parameters presents a significant bottleneck for data analysis.

Purpose of the Study:

• To demonstrate the utility of Artificial Neural Systems (ANS), or neural networks, for processing and analyzing complex flow cytometry data.
• To showcase ANS's ability to overcome the limitations of conventional graphical analysis methods.

Main Methods:

• Application of Artificial Neural Systems (ANS) for simultaneous analysis of multiple flow cytometry parameters.
• Utilizing ANS for classification and real-time analysis of both biomedical (human leukocytes) and oceanographic (marine plankton) data.

Main Results:

• ANS effectively handle nonlinear variations and detect subtle changes in flow cytometry data.
• The systems demonstrated the capability to interpolate and classify cells not included in the training set.
• Real-time classification of six cyanobacteria strains achieved an average accuracy of 98%.

Conclusions:

• Artificial Neural Systems provide a powerful, efficient, and accurate method for analyzing multiparameter flow cytometry data.
• ANS overcome the data bottleneck associated with traditional graphical analysis, offering real-time processing capabilities.