Big Data in Cultural Heritage Imaging Incubator - Day 2
Francesca Gabrieli, National Gallery of Art
| John Delany, National Gallery of Art, demostrates how to collect a dataset on the Roulin’s Baby by Van Gogh, painted in 1888. |
After discussing a number of powerful macro-scale imaging techniques (Day 1 of the Incubator), the Friday discussion focused on methods for micro-scale imaging and mapping, techniques useful for revealing micro-sized heterogeneities and identifying compounds in the outer layers of artwork. Micro-scale mapping can be done both in a non-invasive way (i.e., using micro-scale Raman mapping of small areas of a manuscript) and on physical samples removed from the main piece of art. Different techniques were discussed, with the aim of understanding how they have been used in the past and what their strengths are for future use.
Martin Fisher, from Duke University and one of the Incubator hosts, showed how to get information through nonlinear techniques such as the femtosecond pump-probe spectroscopy, a method of identifying the molecular components of a sample non-destructively. It is a technique that can be used for the understanding of surface degradation of art paint pigments, such as vermillion. Unfortunately, this method currently requires a large optical table and precision alignment, so use in a museum setting is difficult. The current goal is to make such an instrument more compact and rugged.
Wesley Legant, University of North Carolina, presented a high-speed three- dimensional imaging technique named light sheet microscopy. It is currently a technique mainly used for biological applications such as tracking the dynamics of DNA replication and has the ability to follow single molecules. His talk raised many questions about how to exploit such a powerful technique and adapt it for the study of complex cultural heritage materials including the possibility of using it for the 3D Imaging of samples and using it for the study of the dynamics of pigment-binder interaction in a paint layer.
Louisa Smieska, a scientist from the Cornell High Energy Synchrotron, talked about 3D confocal XRF microscopy and explained how it can be used to obtain in-depth characterization of elements and impurities in a cross-sectional area. Karen Trentelman, Getty Institute, and Silvia Centeno, Metropolitan Museum of Art, New York, discussed confocal Raman mapping and described how it is a very promising technique that has not been fully utilized up to now. Raman spectroscopy has high molecular specificity but is extremely challenging to implement since the laser intensity needed can damage or degrade the material under study. The Raman mapping process and instrumentation needs to be upgraded to take advantage of the latest laser sources and optical detectors, and it can be used in a complementary way with other techniques to give a more accurate picture of composition. Pablo Longero, Yale University, showed some examples of details of works of art analyzed with Ramam mapping such as the identification and mapping of challacoloite, probably a degradation product, on a polychromatic terracotta object.
As in the first day, some engineers talked about macro-scale imaging systems that are commercially available. Craig Szymanski, THORLABS, showed some micro-instruments currently available (for multiphoton imaging, confocal imaging, and optical coherence tomography), pointing out their willingness to improve their systems to better meet “cultural heritage field requirements” for the study of heterogeneous surfaces. There was a lively discussion from the audience about the usefulness of micro-scale imaging in addressing some questions, which cannot be answered by macro-scale investigation. However, as pointed out by Koen Janssens, University of Antwert, due to the heterogeneity and complexity of some materials, the very high specificity sometimes can be a disadvantage. Micro-scale imaging should be focused on techniques which have already given good results in this field such as SEM-EDS, ATR-FTIR imaging, micro-Raman, micro-fluorescence mapping, micro-XRF, micro-XRD and relate one to each other. Also, OCT and NMR have been proposed for depth profiling and ELISA (tagging fluorophores with antibodies) for the identification and mapping of binders in a material cross section.
Karen Trentelman pointed out that currently the understanding of the palette for ancient artwork is less tricky thanks to advanced techniques and we can definitely narrow the materials used based on the time period. More interesting topics, both for scientists and conservators, is going towards the “unexpected” – for example, trace elements, minor components, and degradation compounds formed on surfaces. She also made an interesting comment regarding modern art materials, which have been less explored than ancient art, but still require an examination in depth. Modern art has many complicated problems to solve, going from understanding the multiple layers of different binders and pigments in a Rothko painting to fully characterizing the new chemical compounds that have been produced in the last century and that are currently being used by artists.
| Several participants brought posters to further demostrate their work. |
Coming back to the macro-scale imaging, David Saunders, from the British Museum and one of the Incubator hosts, talked about how to identify and map pigments with a multispectral system. In particular, he showed results from UV-induced luminescence imaging and visible light induced luminescence. His study is part of a larger project called LuminArt and is in collaboration with the Courtauld Institute of Art and CNR-nanotec. With their system, they can select a narrow excitation band in the visible and then collect fluorescence in specific narrow bands in the near-IR (800-1000nm) and in the extended near-IR (up to 1700nm), identifying and mapping pigments such as Egyptian blue on paintings and 3D objects. The portable set-up they have created allows them to acquire data in open sites (where it is not possible to block the external light) and in other difficult working conditions.
Discussion about macro-scale imaging was continued by John Delaney who pointed out that the use of hyperspectral rather than multispectral or high-resolution scanning has to be planned carefully depending on what both the scientists and conservators are looking for. Understanding the needs of conservators and archeologists is essential in deciding what type of imaging method to use. The mobility of the imaging system and of the art objects also plays a fundamental role in the choice. A close collaboration with conservators and curators in this field is crucial.
Some final points made during the discussion concerned the processing of the acquired data and the desire for calibration, mosaicking, and registration software to be user friendly. The data has to be manipulated carefully, not only relying on spectral libraries because spectral features can change as materials are intimately mixed. Statistical and chemometric approaches, such as principal component analysis, can be used but it is important to always rely on spectral features for identification!
The last part of the Big Data in Cultural Heritage Imaging Incubator happened behind the scenes of the National Gallery of Art (NGA) museum. Incubator participants and a few OSA staff members were taken into the laboratories of the department of scientific research at the gallery. Barbara Berrie, head of science department at NGA, showed some of the results of her research using mapping SEM-EDS on a cross section of a painting by Giotto. The micro-scale imaging technique allowed her to find a rare carbonate-based mineral called mixite. As mentioned earlier, the finding of the “unexpected” can be of high relevance for the study of a painting’s history or technique. The tour ended in the macro-scale imaging laboratory where John Delaney, Kate Dooley, and Francesca Gabrieli showed how to collect a dataset with a visible hyperspectral camera (400-1000nm) on the painting Roulin’s Baby by Van Gogh, painted in 1888. That painting has been extensively studied to identify and map its paint pigments, as well as to fully understand the fading of a particular “red lake” pigment that the artist used. Delaney showed the reconstruction of the image and the characteristic spectral shapes that allow pigment identification. The NGA lab tours were an exciting highlight of the Incubator!
Thanks to the hosts and the participants, this OSA Incubator was a perfect way to brainstorm about how to improve the quality of the data and instrumentation used in the field of cultural heritage imaging. It opened everyone’s minds to the capability of new instruments to help answer important scientific questions and to lead to better methods of cultural heritage conservation. The hosts are working on a whitepaper to help pull the discussions and recommendations that came from this Incubator together to share with a wider community so stay tuned for more to come!
