2.1.

Optical Coherence Phase Microscopy

In the system used for this study, the light source and the spectrometer were parts of a commercial OCT engine (Callisto, Thorlabs). The superluminescent diode (SLD) is centered around 930 nm with a FWHM bandwidth of 90 nm. The resolution given by the manufacturer is 7 μm in air, corresponding to 5 μm in tissue. The spectrometer recorded 1200 spectra per second.

The light was coupled out of the OCT engine into a single mode fiber and collimated onto a pair of galvanometers for raster scanning. The custom scanning head was attached to an inverted microscope (SPi95, Brunel microscope) and delivered to a 10× objective (NA = 0.17) or a 20× (NA = 0.4).

As such, the system was in a common path set-up [Fig. 1a] without reference arm as in a conventional Michaelson interferometer. Instead, a strong reflection within the sample arm was used as the reference. As a result, there was an increase in phase stability as common noise (i.e., differences of temperature between reference and sample arms, vibrations, fluctuations) was more efficiently rejected.

Fig. 1

(a) Schematic of the optical coherence phase microscope. A superluminescent light source (SLD) was sent to a custom scanning head attached to an inverted microscope. The interferences between a reference reflection and the light backscattered within the sample were measured as a modulation of the spectrum with a spectrometer. The light source and the spectrometer were part of a commercial OCT engine (Thorlabs). (b) Representative B and A scans (c) of a glass coverslip covered with cell culture medium measured by the OCPM. In (b), we can see the top surface of the cover slip appearing as a bright line. This line corresponds to the peak in the A scan (c). The intensity is reported in db.

The backscattered light reflected from within the sample interfered with the reference reflection, generally the bottom surface of a coverslip, perpendicular to the laser beam, onto which the cells were directly cultured. These interferences were traveling along the same path back to the spectrometer. A spectrum was measured at each x,y point of the sample by raster scanning.

In spectral domain OCT, the interferences between the sample and the reference arms were spectrally dispersed and measured at each k with the spectrometer. As a result, we observed a modulation of the light source spectrum. The spectral intensity I(k), a function of the wavenumber k, is given by

The depth information profile, the A scan, originating from the light being scattered at a different depth within the sample, can be retrieved with an inverse Fourier transform. The discrete Fourier transform gives access to the intensity, I(z), and the phase, Φ(z), of the light beam at different depths, z

The intensity, I(z), provides structural information from the sample. The phase, Φ(z) is related to the optical properties at the depth z

Phase stability determines the minimum phase variations above the noise floor that are detectable by the system. Theoretically, the phase variance can be calculated as^{19, 21}

The phase stability was defined as the standard deviation, i.e., the square root of 〈Φ(z)²〉. With a theoretical SNR of 83 db in air, the system had a theoretical phase stability of 7×10⁻⁵ radians. Note that SNR(db) = 10 log 10(SNR).

In the case of 2D imaging, only the top surface was of interest; whereas in 3D structures, the intensity and phase were measured within the depth of field of the objective used.

2.2.

Phase Fluctuations Measurements

We measured phase fluctuations by collecting 50 successive B scans at the same location, yo. We collected 400 A scans in the x direction for 2D imaging and 500 for 3D imaging. At a scanning rate of 1200 s⁻¹, the phase Φ(ti, x, yo, z) = Φi(x, z) was measured every 0.33 s for 2D and 0.41 s for 3D imaging. Respectively, the total acquisition times were of 16.5 s and 20.5 s.

The phase fluctuations were defined as

2.3.

Cell Culture

Human adipose derived stem cells (ADSCs) (Invitrogen) were expanded to passage two in a low serum medium (complete RS MesenPRO) with 2 mM L-glutamine according to the supplier protocol and cultured onto glass coverslips at least one day prior to imaging in a humidified incubator at 37 ° and 5% CO₂.

2.4.

Three-Dimensional Cell Culture

2.5.

Electric Cell-Substrate Impedance Sensing

A commercial ECIS system (ECIS Zθ, Applied Biophysics) was used to demonstrate that ADSCs experienced a continuous micromotion. The time and frequency dependent complex impedance, Z(t, f ), of the electrode-cell interface was recorded at multiple frequencies. Cells were grown on eight-well ECIS arrays containing one 250-μm gold microelectrode per well. Measurements were performed directly in the cell culture medium allowing real-time monitoring, while both the ECIS arrays and the measurement station were kept in an incubator with high-humidity at 37 °C and 5% CO2.

The real part of the impedance, R(t, f ), was normalized and filtered to remove DC components. The fast Fourier transform [R(t, f )], was used to calculate the instantaneous power spectral density (PSD) and plotted on a log-log scale. A micromotion slope was determined by linearly fitting the low frequency part of the PSD.

3.1.

Measured Phase Sensitivity

Figure 1b shows a typical B scan, acquired with the OCPM, of a glass coverslip with cell culture medium on top. The top surface appeared as a bright line in the B scan. The corresponding A scan (in depth profile) at x = 200 is represented in Fig. 1c. In a liquid medium, the SNR is far below the theoretical one and was measured at ∼35 db. Consequently, the measured phase sensitivity was on the order of 0.01 radians, which could be directly translated into a corresponding minimum measurable displacement of 0.56 nm by using Eq. 5. The OCPM was, therefore, able to measure any change in phase without ambiguity between 0.56 and 350 nm (for n = 1.33).

3.2.

Live Cells Display a Constant Micromotion

The impedance of ADSCs cultured on gold microelectrodes [see Fig. 2a] displayed continuous multiscale fluctuations, as shown in Figs. 2b, 2c. The impedance fluctuations were shown not to be due to electrical noise but to “biological noise.” Typically for healthy cells, the slope of the power spectral density was measured between −1.9 and −2.1, which can be compared to a null slope for theoretical white noise or a no cell control. In this study, we measured a slope of −1.9 for ADSCs as shown in Fig. 2d. The slope of the PSD fell to zero for ADSCs fixed in formalin. These results together with the measured phase sensitivity indicated that phase fluctuations corresponding to cellular micromotion could be measured for live ADSCs.

Fig. 2

(a) ADSCs were cultured on top of 250-μm gold microelectrodes (image reproduced from www.appliedbiophysics.com). (b) Recorded resistance of the cell-electrode interface. Fluctuations in the resistance were recorded even after cells reached confluency, demonstrating a constant activity termed as micromotion, and correlated to cell vitality. (c) Fluctuations in the resistance occurred at different time scales. (d) The slope of the PSD of these fluctuations was calculated at –1.9 and was similar to values reported for healthy cells. Cell-free controls had a PSD slope close to zero.

3.3.

Optical Coherence Phase Microscopy Two-Dimensional Imaging of Cells

ADSCs were imaged on top of glass coverslips via OCPM by collecting 100 B scans in the y direction. In this particular case, only the top surface was of interest, as demonstrated in Fig. 1b. At this depth, we reconstructed the intensity map, I(x, y) [Fig. 3a], and the phase map, Φ(x, y) [Fig. 3b]. I(x, y) was expressed in decibels and Φ(x, y) in radians. Φ(x, y) was within [–π, π] and was not unwrapped on this image. The scanning step was set to 1.5 μm in the X direction (A scan) and to 3 μm in the Y direction (B scan). Both maps clearly showed general cell morphology. The resolution can be increased by switching to a 20× objective to reveal intracellular features as described in Ref. 23. However, in this study we were interested in optically averaging the phase fluctuations resulting from cellular micromotion, and we consequently settled for a lower optical resolution.

Fig. 3

(a) Intensity map, I(x, y), of live ADSCs cultured on a glass coverslip. (b) Corresponding phase map, Φ(x, y). The phase was not unwrapped, and the displayed colorbar is in units of [−π π]. Scanning steps [Ascan (x), Bscan (y)] in these pictures were set to 1.5 μm × 3 μm. Corresponding 50-μm scale bars were plotted.

3.4.

Two Dimensional Phase Fluctuations

We collected 50 B scans at the same y location and compared the phase fluctuations for live cells and cells fixed in formalin and stored overnight in PBS. The differences between live cells versus fixed cells were clear when plotting either the standard deviation [Fig. 4a] or the maximum of the discrete derivative of the phase at the same position x [Fig. 4b]. The maximum value has a direct physical meaning since it can be interpreted as a physical displacement, i.e., a vertical micromotion with Eq. 5.

The corresponding displacement scale was plotted on the right of the graph under the approximation that the refractive index is constant. It shows that the OCPM measured a maximum cellular displacement of 6 nm in this recording.

A few smaller phase fluctuations were recorded in cells fixed in formalin. They might be attributed to Brownian motion in the fluid or other sources of phase noises. Note, however, the difference in spatial spreading between phase fluctuations that could be attributed to live cellular micromotion, and the spike appearance of δΦ for fixed cells.

3.5.

Three Dimensional Phase Fluctuations

Figure 5a shows an OCPM B scan of a control scaffold without cells. The porous fibers running in the plane perpendicular to the page can be seen. 200 B scans were collected to reconstruct the sample in 3D. En face imaging [Fig. 5b] at a chosen depth, z, more clearly showed the structure of the fiber. In Fig. 5c, the 3D structure of the cell-free sample was represented with open-source software (V3D).

For 3D phase fluctuation studies of cell-seeded scaffolds, we collected x-z cross-sections to image in depth within the sample. Statistics were performed over 50 cross-sections recorded at the same location yo = 0.

The mean intensities were represented in Figs. 6a, 6b for live and fixed 3D cell culture, respectively. The intensity pictures clearly revealed the fibers of the scaffold, displaying a similar structure to that observed in Fig. 5. In addition, we observed cells growing within the scaffold for both live [Fig. 6a] and fixed cells [Fig. 6b]. Based on the intensity pictures, no differences could be observed between the two conditions. In both cases, the cell-to-scaffold contrast resulted from an a priori knowledge of the cell-free structure.

Fig. 5

(a) OCPM B scan of a blank scaffold. Porous fibers running in the direction perpendicular to the page were visible. (b) Reconstructed en face imaging from the acquisition of 200 B Scans in the y direction. (c) 3D projection of the scan data showing the 3D scaffold structure. Scanning steps [Ascan (x), Bscan (y)] in these pictures were set to 1.5 μm × 3 μm. Axial resolution in z is given by the OCPM axial resolution (i.e., 5 μm per pixel). Corresponding 50-μm scale bars were plotted.

Fig. 6

(a) Intensity B scan for a live 3D cell culture. (b) Intensity B scan for a fixed 3D cell culture. The mean intensity over 50 B scans was plotted in all pictures. No clear contrast existed between the cells and scaffold material. Structurally round hollow fibers could be distinguished from the cells growing between them. Based upon mean intensity values, no differences could be made between live and fixed 3D samples. (c) Standard deviation map of the phase fluctuations for live 3D cell culture calculated over 50 B scans; and (d) δΦ standard deviation map for fixed 3D cell culture. Live cells were characterized by a higher standard deviation of δΦ when compared to fixed cells over the total acquisition time of 20.5 s. As a result, the cell-to-scaffold contrast was increased. Scanning steps in Ascan (x) in these pictures were set to 1.5 μm. Axial resolution in z is given by the OCPM axial resolution (i.e., 5 μm per pixel). Corresponding 50-μm scale bars were plotted.

However, when we plotted the standard deviation of the phase fluctuations, δΦ(x, z), of live [Fig. 6c] and fixed 3D cell cultures [Fig. 6d], we noted striking differences between live and fixed 3D cell culture. Only samples with live cells displayed a standard deviation of δΦ(x, z) above the background noise level. In these samples, the standard deviation of the δΦ(x, z) for the scaffold fibers were at the background noise level and, as a consequence, the cell-to-scaffold contrast was enhanced. No such contrast existed for fixed samples and both cells and scaffold displayed standard deviations under the noise threshold. Note that the phase and its derivatives were not calculated outside areas where the intensity was below the noise threshold. Hence, the image background of zero is artificial. In the phase fluctuations map, several spikes can be observed in live cells, scaffold material, and fixed samples. These spikes could be due to random phase instabilities and/or to phase wrapping creating a large shift in the phase derivative.

Similar to Fig. 4, the phase fluctuations of live cells were spatially spread over defined areas. This could be used advantageously to average over the noise represented by the phase fluctuation spikes.

Fig. 4

(a) Plot of the phase fluctuations, δΦ standard deviation, along x calculated over a total acquisition time of 16.5 s at a 1200 s⁻¹ acquisition rate. (b) Plot of the maximum of the phase fluctuations, δΦ, with the same parameters. Live cells (plain line) were characterized by large cell fluctuations when compared to fixed cells (stars). The corresponding scale in nanometers is displayed on the right-hand side of the plot. Scanning steps [Ascan (x)] in these pictures were set to 1.5 μm.