Imaging systems:
Polychromatic Polarization microscope (PPM) imaging:
The PPM is based on a standard microscope with white light illumination, which is equipped with a special polychromatic polarization state generator and achromatic circular analyzer. The polarization state generator produces polarized light with the polarization ellipse orientation determined by the wavelength, which we call the spectral polarization fan. An example of the fan with the right polarization ellipses for the visible spectrum is shown in Fig. 7. All polarization ellipses have the same ellipticity angle e~40º
If the specimen under investigation isn’t birefringent then the beam passes it without alteration of the polarization. The circular analyzer will evenly, partially transmit all wavelengths, and the output beam will stay white. If the object is birefringent then it modifies the spectral polarization fan. For example, a particle with phase retardation ~10º and the fast axis at 45º will add ~5º to the red polarization ellipticity angle and subtract ~5º from the green ellipticity angle. The red component will have the right circular polarization with ellipticity angle 45º, and it will be extinguished completely by the left circular polarizer. The green component will have an ellipticity angle ~40º, and its transmission will be increased in ~4 times. So, the birefringent particle will be mostly green. If the particle or the spectral polarization fan is rotated by 90º, the picture becomes complementary and the birefringent particle is mostly red.
In order to increase the image contrast and suppress the contribution from non-birefringent structures we can subtract one complementary bright-field PPM image from another. A computed differential PPM image depicts colors that are generated by the birefringence only and eliminates the stain colors. For further analysis we can transform the differential PPM image to monochromatic by using its brightness. Then overlap the monochromatic PPM image with conventional brightfield image captured in unpolarized light. This workflow is summarized below, and it is also illustrated in Fig.8.
Step 1. Take two complementary bright-field PPM images with spectral polarization fans at 0º and 90º.
Step 2. Compute a difference. The differential PPM image will show the non-birefringent structures in black and the birefringent structures in color.
Step 3. Replace the variety of birefringent colors with one monochromatic color, for example by green.
Step 4. Take a conventional bright-field image in unpolarized light.
Step 5. Combine differential PPM image and conventional bright-field image.
A white balance step can be used to correct the camera hue for these images before computing the result. The image processing code in Python, user instructions, and example data are available at https://github.com/uw-loci/polychromatic-polarization.
For the images in Figure 8, we used an inverted light microscope Olympus IX81 (Olympus America, Center Valley, PA, USA) equipped with objective lens UPLFL20xP/NA0.5 and 100W halogen lamp and images were collected using Olympus DP73 color camera. All images were taken in white light without any filter. BF and PPM images in figures 4 to 8 were collected using an upright light microscope Olympus BX60 (Olympus America, Center Valley, PA, USA) equipped with objective lens 10x/NA0.3, 20x/NA0.5 and 40x/N0.75 and 100W halogen lamp and the images were collected using a Olympus DP25 Color CCD camera. Figure 3 was done using data from both systems.
Second Harmonic Generation (SHG) imaging: All the SHG imaging in this study was done with a custom built integrated SHG/bright field imaging system. A MIRA 900 Ti: Sapphire laser (Coherent, Santa Clara, CA) tuned to 780 nm, with a pulse length of less than 200 fs, was directed through a Pockels cell (ConOptics, Danbury, CT, USA), half and quarter waveplates (ThorLabs, Newton, NJ, USA), beam expander (ThorLabs), a 3 mm galvanometer driven mirror pair (Cambridge, Bedford, MA), a scan/tube lens pair (ThorLabs), through a dichroic beam splitter (Semrock, Rochester, NY) and focused by 20X/0.75NA air objective lens (Nikon, Melville, NY). SHG light was collected in the forward direction with a 1.25 NA Abbe condenser (Olympus) and filtered with an interference filter centered at 390 nm with a full width at half maximum bandwidth of 18 nm (ThorLabs MF390-18). The back aperture of the condenser lens was imaged onto the 5 mm aperture of a H7422-40P GaAsP photomultiplier tube (Hamamatsu, Hamamatsu, Japan) the signal from which was amplified with a C7319 integrating amplifier (Hamamatsu) and sampled with an analog to digital converter (Innovative Integration, Simi Valley, CA). Timing between the galvo scanners, signal acquisition, and motorized stage positioning was achieved using our custom software called WiscScan. SHG images were tiled with 5% overlap using automation provided by WiscScan. Stage positions for individual images and pixel size data were read in by Bio-Formats 58 image metadata and this was then used by the grid/collection stitching ImageJ plugin 59 to reassemble a high-resolution large field of view image of the entire imaged area.
Histological samples:
For this study we have retrospectively used two well studied tissue microarrays containing breast and pancreatic cancer. Complete description for breast samples can be found in 1 which demonstrated the prognostic value of the TACS3 in human patient; and for pancreatic samples in 15, which has shown that collagen alignment is a negatively prognostic factor in pancreatic ductal adenocarcinoma progression. However, to briefly describe the samples, all tissues were formalin-fixed and paraffin-embedded, then cut into 5 µm thin slices, affixed to a slide and stained with hematoxylin and eosin (H&E) before mounting with a coverslip.