Automated Flow Cytometry Analysis
Flow Cytometry (FC) is a powerful technique for the analysis of multiple parameters related to the physical and biochemical characteristics of single cells or particles. It allows the study of many characteristics in minimum time. FC has a wide range of applications, in diagnosis and clinical monitoring of hematological diseases (leukemias- lymphomas), infectious diseases (HIV/AIDS, tuberculosis, malaria, etc.), immonodefieciencies, and reproductive disorders (absolute sperm count, viability and sperm quality). FC is a very useful tool in cancer research, offering the opportunity to study different cell types and cell lines under different conditions regarding a wide range of parameters (e.g. DNA analysis, apoptosis, cell cycle analysis, cell proliferation, cytokine production, viability).
A key step in flow cytometric analysis is to focus on particular groups (clusters) of cells, widely known as ”gating”, which is accomplished through visual inspection of plots and is highly influenced by the user’s experience. These groups of cells are then identified by their unique characteristics. The traditional gating procedure is time consuming and subjective. This, combined with the fact that the FC technology is rapidly evolving, prompted us to develop an automated analysis method for FC data.
Gating corresponds to a data clustering problem in computer science, by assuming that distinct populations of cells form distinctive clusters. As a result, traditional clustering approaches have been used, such as the k-means algorithm and some of its variations, fuzzy k-means, k- medoids and Gath Geva. INAB/CERTH in collaboration with ITI/CERTH are in the process of developing a FC analysis software to fully automate the entire flow cytometry analysis of FC data from Chronic Lymphocytic Leukemia (CLL) patients.
Using established techniques combined with post-processing methods and algorithms that simulate experts’ decisions, our goal is to produce an objective result that provides an automatic answer, whether a sample is normal or pathologic. The method is validated comparing the experts’ manual analysis against the automated analysis with satisfying results.
The main aim of the project is to create a FC analysis program that produces objective results with minimum user interference, and the expansion of its application to other diseases besides CLL.