Hahn, HorstHöfener, HenningHenningHöfener2020-04-282020-04-282019-09-02https://media.suub.uni-bremen.de/handle/elib/427010.26092/elib/54Examination of tissue in pathology plays a central role in many diseases, including most cancers. Pathologists are remarkably good at conducting qualitative investigations, including finding and understanding different tissue patterns and textures. However, quantitative examinations, which are mostly required for the assessment of cellular structures, contain large inter- and intra-observer variability. Automated quantification of cellular structures using digitized histological tissue sections promises to improve accuracy, reproducibility and efficiency of quantitative assessments. However, histological images exhibit large variability, artifacts and clustered structures, which presents a challenge for automated analysis. This cumulative dissertation aims at bringing the automated quantification of cellular structures closer to practical applicability. To this end, efficient analyses will be developed that are optimized with regard to these challenges.enDigital PathologyHistologyImage AnalysisDeep learningMachine LearningNuclei DetectionBiomarker Quantification0Dissertationurn:nbn:de:gbv:46-00108607-13