Abstract: Last century the field of heritage sciences expanded beyond imagination. The inventions of X-ray radiography and infrared reflectography allowed experts to investigate paintings below the surface as well. More recent developments led to the advent of the field of hyperspectral imaging, to which the advanced chemical imaging methods, used in this thesis work, belong. These techniques not only allow to identify the components present in artworks, but also to visualize their distribution over these objects. The resulting distribution maps permit a broader public to interpret the scientific data and to relate these results with the artwork itself. During this thesis work a range of flat artworks were investigated in a non-destructive manner using mainly two macroscopic imaging techniques: macroscopic X-ray fluorescence scanning and macroscopic Fourier transformed mid-infrared scanning in reflectance mode. The resulting images were sometimes supplemented with microscopic techniques on a minute selection of samples to fully understand the layer build-up, composition and distribution of these materials over the stratigraphy. Illuminated manuscripts pushed the interpretation of the macroscopic imaging techniques: due to the impossibility of sampling, all answers had to be obtained non-destructively. Documenting masterpieces such as the Ghent Altarpiece by means of chemical imaging techniques, helped the restoration team, assisted by the international commission to make the daring decision of manually removing the non-original paint layers. Scanning stained-glass windows allowed experts to document the panels, create situation reports, identify later infills and guide the restoration process in a more efficient manner. By initially applying non-destructive imaging techniques, many of the research/conservation questions could already be answered. Based on the resulting distribution maps, only a very limited amount of sampling was required to obtain a representative set to answer the remaining questions. In most cases the combination of multiple methods was necessary to fully understand the situation. A similar trend could be seen in the research field: the collaboration between divergent disciples was often required in order to explain all observations. In order to completely break through, the scanning speed of these techniques has to increase even more in order to cover an acceptable surface in one workday. Parallel with the operational speed, the (basic) data treatment should also be streamlined more in order to allow a broader user group to access the results. Once these two improvements are carried out, these techniques become accessible to a larger public.