| This project is initiated by Marius Nap of the Pathology Department of the Atrium Hospital in Heerlen in collaboration with Gilbert van Zeijl of MEMIC. The objective of the project is to enhance the analysis and classification of tumour cells on the basis of their DNA contents. Hitherto, the analysis is mainly performed by human visual analysis of DNA histograms obtained from a flowcytometer. Pathologists have expertise in interpreting the histograms but their interpretations may differ depending on prior experience and laboratory setting. Therefore, support from automatic analysis techniques is needed. A dataset of carefully labelled DNA histograms was created. An automatic classification procedure was developed that combines pre-processing with adaptive classification. Throughout the year, many new insights were obtained on the nature of DNA histograms. More importantly, possible sources of error have been identified which may lead to improved classification procedures for DNA analysis. A paper describing the approach and some initial results has been prepared and will be submitted to a medical journal. In the project, we collaborate with Bob Duin of the pattern-recognition group of the Technical University Delft. |
