Real-time adaptive clustering of flow cytometric data

摘要

In dealing with massive flow cytometric data, a real-time adaptive clustering technique referred to as RTAC has been developed. This technique adopts the brain metaphor for information processing. The problem-solving structure is configured as a connectionist network. Information is encoded in the form of connection weights. The structure evolves as more data are seen by adjusting its weights, governed by a learning equation. The number of clusters need not be predefined. The algorithm is fast and robust. The results are reported from the domain of measuring the antigenic properties of blood samples. Its relations to other clustering alternatives are discussed. The technique has been validated statistically with respect to self-consistency.