Validation of A Computational B-Cell Lymphoma Analysis by Flow Cytometry Data

摘要

Flow cytometry is a valuable ancillary study to aid in the diagnosis of hematolymphoid malignancies, including B-cell lymphomas. Although there are some computerized analysis available to process the flow cytometry data sets, most of the current automatic technique has not been fully developed. To provide a solution of diagnosing B-cell lymphomas, a prototype of automatic system was presented in a previous paper with testing limited to available data. Recently, two newly collected flow cytometry data sets were available to allow us to further validate and extend the work. In particular, one of the two data sets (minimal residual disease data set) consists of subjects with very fewer B-cell lymphomas cells, which greatly increases the difficulty of the automatic diagnosis. To strength our system, a confidence level of diagnosis is defined to assist the physician making a better decision. In addition, the validity of the system is presented by the strong correlation between the diagnosis of the automatic system and the conventional manual process.