2.1.

Setup for Digital Holographic Microscopy

Figure 1 depicts the schematic setup of the applied digital holographic microscopy system. An inverse microscope arrangement, in combination with a transmitting light illumination, enables the investigation of transparent samples in a liquid-filled Petri dish, such as living cells in culture medium. The emitted light of a frequency-doubled $\mathrm{Nd}:\mathrm{YAG}$ laser ( $\lambda =532\mathrm{nm}$ , $P\approx 10\mathrm{mW}$ ) is divided into an object illumination wave $O$ and a reference wave $R$ . A condensing lens is utilized for an optimized illumination of the sample as for conventional white light illumination. The reference wave is guided directly by a beamsplitter to a charge-coupled device (CCD) sensor with IEEE1394 interface (Sony XCD-X700, $8\mathrm{bit}$ , $1024\times 768\mathrm{pixels}$ , pixel pitch $6.25\times 6.25\mathrm{\mu }\mathrm{m}$ ) that is applied for the digitalization of the holograms in the hologram plane HP at $z=z_0$ . Holographic off-axis geometry is realized by a slight tilt of the reference wavefront with the beamsplitter against the wavefront of the object wave. To enhance the system's lateral resolution that is restricted by the pixel pitch of the applied CCD sensor, the transmitted object wave is magnified by a microscope lens. The magnification is chosen in such a way that the smallest imaged structures, given by the restriction of the Abbe criterion, are oversampled by the image recording device (microscope lenses: Zeiss Acroplan LD $20\times$ , $\mathrm{NA}=0.4$ , Zeiss Acroplan LD $40\times$ , $\mathrm{NA}=0.6$ ). In this way, the maximum diffraction-limited resolution of the optical imaging system is not decreased by the numerical reconstruction algorithm described in Sec. 2.2.

Fig. 1

Schematic setup for an inverse off-axis digital holographic microscopy system with transmitting light illumination of the sample (for explanation of $HP$ , $z_0$ , $z_{IP}$ , and $\mathrm{\Delta }z$ , see text).

2.2.

Quantitative Holographic Phase Contrast Measurement

The reconstruction of the digitally captured holograms is performed by the application of a nondiffractive reconstruction method (NDRM). The intensity distribution $I_{HP}(x,y,z_0)$ in the hologram plane HP, located at $z=z_0$ , is formed by the interference of the object wave $O(x,y,z=z_0)$ and the reference wave $R(x,y,z=z_0)$ :

The parameters $K_x,K_y,L_x$ , and $L_y$ in Eq. 2 cannot be obtained directly from the geometry of the experimental setup with an adequate accuracy, and for this reason are adapted once before the measurements by an iterative fitting process in an area of the hologram without sample (for a detailed description, see Ref. 9).

Digital holographic phase contrast microscopy requires, in correspondence to microscopy with white light illumination, a sharply focused image of the sample. For the case that the object is not imaged sharply in the hologram plane HP during the hologram recording process, e.g., due to mechanical instability of the experimental setup or thermal effects, in a second evaluation step a further propagation of the object wave to the image plane can be carried out for subsequent focus correction. The propagation of $O(x,y,z_0)$ to the image plane $z_{IP}$ that is located at $z_{IP}=z_0+\mathrm{\Delta }z$ in the distance $\mathrm{\Delta }z$ to HP can be carried out by a Fresnel transformation^{9, 10, 23} or, as in this contribution, by a convolution algorithm^{24, 25}:

From $O(x,y,z_{IP})$ , in addition to the absolute amplitude $\mid O(x,y,z_{IP})\mid$ that represents the image of the sample, the phase information $\mathrm{\Delta }{\phi }_S(x,y,z_0)$ of the sample is reconstructed simultaneously:

2.3.

Refractive Index Determination and Shape Measurement of Living Cells

The reconstructed object wave phase information ${\varphi }_O(x,y,z_{IP})$ obtained from Eq. 4 can be applied under suitable conditions for the integral refractive index determination of semitransparent samples, such as cells in culture medium. Figure 2 shows the scheme of an experimental setup for the measurement of the refractive index of cells by digital holographic microscopy. The cells are covered with a capping glass that is pressed onto the sample in such a way that the distance $d$ to the carrier glass is nearly constant in the investigated image field. In this case, the phase change $\mathrm{\Delta }{\phi }_{\mathrm{cell}}$ affected by a cellular sample in comparison to the culture medium with the known homogenous refractive index $n_{\mathrm{medium}}$ is:

Fig. 2

Experimental setup for refractive index determination of cellular specimen. $d$ is the distance between the glass capping and carrier slide.

2.4.

Cell Culture

The investigated human pancreatic ductal adenocarcinoma cell lines PaTu 8988S and PaTu 8988T were obtained from the German Collection of Microorganisms and Cell Cultures (DSMZ, Braunschweig, Germany). Both cell lines were established in 1985 from a liver metastasis of a primary pancreatic adenocarcinoma from a 64-year-old woman. PaTu 8988S represents a highly differentiated carcinoma cell line, PaTu 8988T a poorly differentiated adenocarcinoma with a high metastatic potential.²⁶ PaTu 8988T cells were retrovirally transduced with an E-cadherin expression construct containing an E-cadherin cDNA in the expression vector pLXIN (Clontech, Palo Alto, California). Cells were selected, cloned, and analyzed for E-Cadherin expression.²⁷ The morphology of the cell lines was examined by scanning electron microscopy (SEM).

Pancreas tumor cell lines were cultured in Dulbecco Modified Eagle Medium (DMEM) supplemented with 5% fetal calf serum (FCS), 5% horse serum, and 2-mM L-glutamine at 5% ${\mathrm{CO}}_2$ . For the digital holographic investigations, the cells were trypsinized, seeded subconfluent on glass slides or tissue culture plates, cultured for $24\mathrm{h}$ , and analyzed at room temperature and normal atmosphere.

For the dynamic measurements, cells were incubated with $10\mathrm{\mu }\mathrm{M}$ of the cell permeable actin cytoskeleton disrupting toxin Latrunculin B (Calbiochem, San Diego, California).²⁸

3.1.

Refractive Index Determination of PaTu 8988 Cells

In a first experimental step, the refractive index of the three pancreas tumor cell lines is determined. Therefore, adherently grown cells (PaTu 8988S, PaTu 8988T, and PaTu 8988T pLXIN E-Cadherin) on a glass carrier are covered with a capping glass as described in Sec. 2.3. Afterward, holographic phase contrast images are obtained from the samples by application of the experimental setups depicted in Figs. 1 and 2. Figure 3 shows an exemplary result of an investigated PaTu 8988T cell for refractive index determination that also illustrates the evaluation process of the measurement data. Figure 3a depicts the region of interest $(451\times 451\mathrm{pixels})$ from a captured digital hologram of the PaTu 8988T cell that is located near enough to included locked air. Figure 3b shows the reconstructed holographic amplitude image, while Fig. 3c presents the corresponding wrapped quantitative phase contrast image of the PaTu 8988T cell modulo $2\pi$ . For imaging, both amplitude and phase are transformed to 256 gray levels. Figure 3d shows the plot of an unwrapped cross section through the phase distribution along the dashed white line in Fig. 3c. As a result, $\mathrm{\Delta }{\phi }_{\mathrm{air}}=5.02\pm 0.25\mathrm{rad}$ and $\mathrm{\Delta }{\phi }_{\mathrm{cell}}=0.94\pm 0.25\mathrm{rad}$ are determined. With the refractive index of the culture medium $n_{\mathrm{medium}}=1.337\pm 0.001$ (obtained by an Abbe refractometer), $n_{\mathrm{air}}=1.000292$ ,²⁹ and $\lambda =532\mathrm{nm}$ , by application of Eqs. 6, 7 a distance between the capping and the carrier glass of $d=1.26\pm 0.063\mathrm{\mu }\mathrm{m}$ and $n_{\mathrm{cell}}=1.40\pm 0.016$ is obtained. Figure 4 shows in comparison the wrapped quantitative phase contrast image modulo $2\pi$ of the three different cell types in 256 gray-level representation [Fig. 4a PaTu 8988S, Fig. 4b PaTu 8988T, and Fig. 4c PaTu 8988T pLXIN E-Cadherin). For noise suppression, the averaged phase data obtained in the white marked areas are inserted into Eq. 6 to calculate the refractive index values. Table 1 shows the results for $n_{\mathrm{cell}}$ obtained from the investigation of 12 single cells (four of each cell type), as shown in Fig. 4. Within the error obtained from the standard deviation, no difference between the values for $n_{\mathrm{cell}}$ of the three cell types is detected. For this reason, the further data evaluation in Sec. 3.2 is carried out for all pancreas tumor cells by applying the average value ${\overline{n}}_{\mathrm{cell}}=1.38\pm 0.015$ obtained from all measured refractive index values.

Fig. 3

Refractive index determination of living pancreas cells by digital holographic microscopy: (a) digitally captured hologram of a PaTu 8988T cell; (b) holographic amplitude image reconstructed from (a); (c) wrapped phase contrast image modulo $2\pi$ reconstructed from (a); (d) cross section through the unwrapped phase contrast image along the dashed white line in (c).

Fig. 4

Refractive index determination of pancreas tumor cells: (a) unwrapped phase contrast image of PaTu 8988S cells (256 gray-level representation); (b) unwrapped phase contrast image of PaTu 8988T cells; and (c) unwrapped phase contrast image of PaTu 8988T pLXIN E-Cadherin cells. White marked areas are the phase data evaluated for the refractive index determination (A is air, C is cell, and M is the culture medium).

Table 1

Integral refractive index for PaTu 8988T, PaTu 8988T pLXIN E-Cadherin, and PaTu 8988S determined by digital holographic phase contrast microscopy.

3.2.

Shape Measurement of Living PaTu 8988T Cells

In the next experimental step, the information about the integral refractive index data obtained from Table 1 is applied to determine the thickness of adherently grown living PaTu 8988T cells in a Petri dish with cell culture medium. For the experiments, holograms of adherent PaTu 8988T cells and PaTu 8988T pLXIN E-Cadherin cells are captured in the transmitting light arrangement (see Fig. 1). From the determined phase data that represent the optical path length change affected by the samples, the thickness $d_{\mathrm{cell}}$ of the cells is obtained by application of Eq. 8 and the average refractive index ${\overline{n}}_{\mathrm{cell}}=1.38$ data from Table 1. Figure 5 depicts results obtained on a PaTu 8988T cell. Figures 5a and 5b show the reconstructed holographic amplitude image and the quantitative phase contrast image modulo $2\pi$ in gray-level representation. Figure 5c depicts the unwrapped phase data with included gray-level legend for the phase and the corresponding calculated cell thickness. Figure 5d shows a rendered pseudo 3-D representation of the cell thickness in correspondence to the lateral cell size in comparison to a SEM image of a PaTu 8988T cell [Fig. 5e]. The SEM electron micrographs were taken by a LEO 1530 VP (LEO Electron Microscopy Group Ltd., Cambridge, United Kingdom) in high vacuum mode $\left({10}^{-6}\mathrm{mbar}\right)$ . During all measurements, the acceleration voltage was $5\mathrm{kV}$ and an Everhart Thornley secondary electron detector was used. To prevent artefacts caused by charging of the samples, their surfaces were coated with a thin metal layer followed by a carbon layer. Figure 6 represents the results obtained for two PaTu 8988T pLXIN E-Cadherin cells. For each cell type, the obtained data for the cell thickness were found in good agreement with the appearance of the SEM images. In Fig. 7 , the phase values as well as the cell thickness are plotted for both cell lines corresponding to the cross sections marked by white lines in Figs. 5c and 6c. The different cell types can be differentiated by the phase difference $\mathrm{\Delta }{\phi }_{\mathrm{cell}}$ as well as by the calculated thickness. For the PaTu 8988T cell, a maximum thickness of $d_1=(23\pm 1)\mathrm{\mu }\mathrm{m}$ , and for PaTu 8988T pLXIN E-Cadherin $d_2=(7\pm 1)\mathrm{\mu }\mathrm{m}$ , is obtained. The maximum error for $d_1$ and $d_2$ was estimated by taking into account the spatial phase noise of the phase distributions that was determined to an average value of $0.26\mathrm{rad}$ in the area of the sample. In comparison, for areas without cells, the phase noise is determined to $0.08\mathrm{rad}$ . Assuming that the investigated cells were grown fully adherently on the carrier glass and taking into account the appearance of the pancreas cells in the SEM images, the results for the cell thickness in Fig. 7 represent in good estimation the shape of the cells.

Fig. 5

Thickness determination of living PaTu 8988T cells by digital holographic microscopy: (a) reconstructed holographic amplitude image; (b) corresponding quantitative phase contrast image modulo $2\pi$ ; (c) unwrapped quantitative phase contrast image (for the plot of the cross section along the white line, see Fig. 7); (d) rendered pseudo 3-D plot of the cell thickness obtained from (c); and (e) SEM image of a PaTu 8988T cell.

Fig. 6

Thickness determination of living PaTu 8988T pLXIN E-Cadherin pancreas cells by digital holographic microscopy: (a) reconstructed holographic amplitude image of two PaTu 8988T cells; (b) corresponding quantitative phase contrast image modulo $2\pi$ ; (c) unwrapped quantitative phase contrast image (for the plot of the cross section along the white dashed lines, see Fig. 7); (d) rendered pseudo 3-D plot of the cell thickness obtained from (c); and (e) SEM image of a PaTu 8988T pLXIN E-Cadherin cell.

Fig. 7

Thickness comparison of PaTu 8988T cells and PaTu 8988T pLXIN E-Cadherin cells along the cross section lines in Figs. 5c and 6c.

3.3.

Dynamic Measurements on a Living Human Pancreas Carcinoma Cell (PaTu 8988S)

In a further experimental step, the digital holographic microscopy setup is applied for the detection of dynamic morphological changes of living cells in cell culture medium. Therefore, a hologram series of a single PaTu 8988S cell with a time delay between two recordings of $\mathrm{\Delta }t=10\mathrm{s}$ is captured. At the beginning of the experiment at $t=0\mathrm{s}$ , the marine cell toxin Latrunculin B was added to a final concentration of $10\mathrm{\mu }\mathrm{M}$ to the cell culture medium to destruct the actin filaments of the cell's cytoskeleton. Figure 8 shows the rendered pseudo 3-D plot of the cell thickness calculated from the phase distribution at $t=0\mathrm{s}$ by taking into account the average refractive index from Table 1 [Fig. 8a] in comparison with a corresponding SEM image of a PaTu 8988S cell [Fig. 8b], which shows a similar appearance. The upper row of Fig. 9 represents the wrapped phase distribution modulo $2\pi$ after $t=0\mathrm{s}$ [Fig. 8a], $t=30\mathrm{s}$ [Fig. 8b], $t=80\mathrm{s}$ [Fig. 8c], and $t=180\mathrm{s}$ [Fig. 8d]. With increasing time after the addition of Latrunculin B, the phase distributions indicate the cell collapse affected by the destruction of the cytoskeleton. In the lower row of Fig. 9, the corresponding unwrapped phase data along the marked horizontal cross sections of the distribution (see white arrows in the upper row of Fig. 9) is plotted as well as the cell thickness that is calculated for the assumption of negligible refractive index changes of the cell. During the collapse, $\mathrm{\Delta }{\phi }_{\mathrm{cell}}$ as well as the cell thickness are locally decreased up to about 50%. Furthermore, the crinkle structure of the shrunken cell is clearly visualized in the phase contrast images as well as in the plotted cross sections.

Fig. 8

Dynamic analysis of human pancreatic carcinoma cells: (a) rendered pseudo 3-D plot of the cell thickness obtained from the holographic phase contrast image of a PaTu 8988S cell at $t=0\mathrm{s}$ ; and (b) SEM image of a PaTu 8988S cell.

Fig. 9

Analysis of a living PaTu 8988S cell after addition of a marine toxin (Latrunculin B) to the cell culture medium. Unwrapped phase values and cell thickness along the cross sections (lower row) through the reconstructed quantitative phase contrast images modulo $2\pi$ (upper row) after (a) $t=0\mathrm{s}$ , (b) $t=60\mathrm{s}$ , (c) $t=90\mathrm{s}$ , and (d) $t=190\mathrm{s}$ after the addition of Latrunculin B.