Feature extraction, selection and classifier design in automated time-lapse fluorescence microscope image analysis

Using automated time-lapse fluorescence microscope imaging to measure drug response, e.g. using taxanes and the vinca alkaloids to cancers, has become a common practice in cancer research ands drug discovery. A mature system usually consists of the following subsystems: image acquisition subsystems, image pre-processing subsystems, image analysis subsystems, database subsystems and data analysis subsystems. Particularly, the image analysis subsystems play a key role in determining the overall performance of this system. The image analysis subsystems comprise of a dynamic cellular segmentation module, a feature extraction, a cell-cycle phase identification module, a tracking module and a descriptor quantification module. In this chapter, we will systematically introduce the feature extraction/selection and cell-cycle phase identification module. To illustrate the general techniques mentioned above, an applied laboratorial system which includes the many descriptors representing nuclei, feature selections techniques and online-SVM will be introduced. These techniques can easily be applied for users’ own research purposes.