Navigation auf uzh.ch
Advanced neuroimaging is capable to provide very detailed information on various tissue characteristics of brain tumors, surrounding parenchyma and normal brain including physiological and functional measures. Innovations in medical image analysis are complementary by segmenting various tissue components automatically and by computing multiple and quantitative imaging parameters that are mostly undetectable to the “human eye”, but free of subjective bias such as intra- or inter-reader variability. Computer algorithms for machine learning (ML) and artificial intelligence (AI) have now entered this arena with the potential to further improve imaging diagnostics and to develop novel expert reading and support systems.Currently, however, we are still facing a gap between the innovative technologies available for medical imaging and the clinical imaging routines established which mostly lack optimized and standardized protocols for imaging and data evaluation as well as for the definition of (quantitative) imaging target parameters. Clinical imaging is therefore highly variable within and between institutions, comparability and reproducibility is limited. Furthermore, radiological reporting is traditionally narrative (descriptive), thus subjective, not structured and rarely contains quantitative data. The large pool of existing patient and imaging data is difficult to search and access in a structured manner within RIS/PACS systems. As a consequence technical and practical solutions for these topics are warranted to translate new imaging technologies into clinical applications