Tomographic Phase Microscopy (TPM) - Biomedical Optical Phase Microscopy and Nanoscopy - page 238

( Figure 12.9A E ) were taken for comparison by moving the objective lens at the same

height as Figure 12.9F J , respectively. It is clear that the structures in both of the refractive

index tomograms are well matched to the brightfield images. However, if we compare the
details of the images at the tomogram slices 1.7 m above the focus, the difference between

tomograms can be seen. Figure 12.9E, J, and N is the zoom-in images of the black
rectangles in Figure 12.9B, G, and L , respectively. Compared with the brightfield image,
tomograms processed using the inverse Radon transform show blurring of fine structures in
the cytoplasm. In contrast, in the tomogram processed by diffraction tomography, the two
lines that separate two nuclei are clearly resolved ( Figure 12.9N ). This demonstrates that
the Rytov approximation is valid for taking the effect of diffraction into account in
reconstructing 3D refractive index maps of live biological cells. As a result, we could
clearly image the details of the 3D structures of a single live cell throughout its entire
volume as well as quantify the refractive index of subcellular organelles.

μ

12.4 Video-Rate TPM

TPM is beneficial in 3D imaging with minimal assumptions. But its data acquisition is slow
because large numbers of independent 2D images are to be recorded to fill the 3D object
spectrum. This fundamentally impedes studying fast dynamics in biological cells. A single
TPM tomogram requires about 10 s of data acquisition [13] . Improving the speed of
tomographic imaging will open up new possibilities in imaging rapidly changing, moving,
or flowing cells. The acquisition time of TPM has been limited by two factors. First, for
optimum image quality, phase images must be acquired at approximately 100 illumination
angles; each phase image requires the capture of four raw frames due to the use of phase-
shifting interferometry, for a total of 400 images per tomogram. Second, the galvanometer
controlling sample illumination angle must be held constant during the acquisition of the
four frames, requiring a settling time of approximately 100 ms after each change in angle.
In order to overcome these two limitations, a spatial fringe pattern demodulation technique
[29,30] can be applied for individual phase recording in TPM. The method uses only 150
raw images per tomogram and does not require galvanometer settling time. As a result,
full 3D tomograms can be acquired at a rate of 30 Hz [25] .

12.4.1 Experimental Scheme

The setup ( Figure 12.10 ) resembles the TPM system described in Section 12.3 without acousto-
optic frequency shifting in the reference arm. A helium neon laser beam ( λ 5 632.8 nm) is
divided into sample and reference arm paths by a beamsplitter. In the sample arm, the beam
reflected from a galvanometer-mounted mirror (HS-15, Nutfield Technology). A lens
(L1, f 5 250 mm) is used to focus the beam at the back focal plane of an oil-immersion
condenser lens (Nikon 1.4 NA), which recollimates the beam to a diameter of approximately