Optical Coherence Tomography (OCT) is a non-contact and non-destructive optical imaging modality which can visualize and reconstruct the spatial structure of objects. It is well-suited for examining fragile works of art which often have fine layered structure, possess unique historical value, and inherently require safe analysis methods to avoid damage.

Recently, researchers at MIT working with the Harvard University Art Museums and Thorlabs successfully applied swept source OCT using a Fourier domain mode locked (FDML) laser for high-speed, three dimensional OCT (3D-OCT) imaging of gold punchwork in Renaissance panel paintings. The FDML laser was constructed at MIT and had a sweep repetition rate of 42,000 sweeps/sec, with center wavelength near 1300 nm. The laser provided an OCT axial imaging resolution of ~6 µm in varnish or paint. With ~10 mW of power incident on the sample an imaging sensitivity of 100dB was achieved. As an imaging platform, a modified OCT microscope (Thorlabs OCM1300SS) was reconfigured for imaging vertically mounted objects. The beam spot size was ~30 µm, which defines the transverse image resolution. The working distance of the microscope was ~3 cm. A photograph of the experimental setup is shown in Figure 1.

Figure 1. Experimental setup. T – imaging platform: modified OCT microscope (Thorlabs, Inc.); L – light source: Fourier Domain Mode Locked (FDML) laser constructed at MIT; S – Sample.

Researchers imaged punch marks in two different regions of the “Coronation of the Virgin”, a painting on wooden support shown in Figure 1, created by the Master of the Orcagnesque Misericordia (active between 1375 and 1400AD). The photographs of the painting are shown in Figure 2. The rectangles A and B in enlarged views indicate the locations of specific punch marks imaged with the 3D-OCT instrument. The three dimensional data sets acquired in these regions consist of 800 x 800 x 512 pixels in horizontal (X), vertical (Y) and depth (or axial; Z) directions. Figure 2 shows details of imaged punch marks.

Figure 2. OCT imaging of gold punchwork: projection images. Left column: enlarged views of gold punchwork photographs. Middle column: projection images obtained by axial summation of cross-sectional OCT images as indicated by yellow arrows in OCT B-scans (right column); false color scale was selected to enhance the contrast of punchwork; size of the images is 4 mm2. Right column: example OCT cross-sectional images selected from the three dimensional data sets; arrow 1 indicates area covered with the red paint, arrow 2 points at a punch marks filled with varnish, arrow 3 – gold foil.

High-speed 3D-OCT enables high density scanning of selected regions of the object in reasonably short times. The high definition data can be used to generate three dimensional virtual models of the punchwork, as shown in Figure 3. This visualization enables intuitive assessment of spatial distribution of punches, their shapes and depths.

Figure 3. OCT imaging of gold punchwork: volume rendering 3D-OCT data sets are used to generate virtual spatial models of imaged gold punchwork. These models can be viewed from any side for intuitive assessment of shapes and depths of punch marks.

3D-OCT instruments are well-suited for applications in art conservation study. The infrared light used in OCT imaging has the ability to penetrate through different materials used for creating works of art. This feasibility study shows that 3D-OCT can be used for examination of paintings containing gold punchwork. Other promising applications of OCT in art conservation include optical stratigraphy of paintings, imaging of underdrawings, glaze and varnish layers, analysis of paint transparency, monitoring of controlled removal of varnish, and analyzing canvas deformation.

More detailed information on this study can be found in a paper accepted for publication in Optics Express: “Comparison of Three-Dimensional Optical Coherence Tomography and High Resolution Photography for Art Conservation Studies”, Desmond C. Adler, Jens Stenger, Iwona Gorczynska, Henry Lie, Teri Hensick, Ron Spronk, Stephan Wolohojian, Narayan Khandekar, James Y. Jiang, Scott Barry, Alex E. Cable, Robert Huber and James G. Fujimoto.