Method of classifying cell nuclei using binary boosted classifiers

摘要

The invention relates to a system for classifying objects in biological specimens. Images of biological tissue are captured using a microscope 3 connected to a camera 1. These images are then segmented and classified using a supervised learning method to identify different nuclei types and to reject artefacts that arise from the segmentation process. The method employs two binary classifiers with boosting. The classifiers are arranged in a cascade system. The classification of different types of cell nuclei and the rejection of artefacts is thus achieved in a completely automatic way. The nuclei may be characterised by parameters such as dimensions, shape, area, Hu moments, jaggedness, and number of grey levels in the image. The boosting method may be Adaboost or Gentleboost. The classifier may be first trained using a training database.