4 Tomographic Diffractive Microscopy …
89
Demodulation is easily performed in Fourier space, in which the different orders of
diffraction are spatially separated. This technique has the advantage, compared to
phase-shifting holography, that a single hologram is enough to extract amplitude and
phase of the diffracted beam, but at the price of a reduced field of view, because of
the demodulation constraints in Fourier space.
Note that combined approached can also be used [52]. For more details about
digital holographic data acquisition and reconstruction, the interested reader will
find extensive description in [53–58].
Indeed, DHM is a very powerful technique of investigation when considering
integral measurements, by projections of the accumulated phase changes when light
travels through the observed samples [59], rapid phase variations, for example, corresponding to samples fluctuations [60, 61], very weak sample phase changes, thanks
to its high sensitivity, which can be linked to other physical quantities, for example,
allowing for optical patch clamp measurements [62]. It is also able to acquire long
time-lapse, without inducing photodamages to the sample, thanks to the low level
of light necessary to acquire interferograms (compared to confocal fluorescence),
which reveals useful for cell cycle, cell division, cell migration, drug influence or
cell apoptosis studies, cell detection and identification [63–66].
However, its main drawback is probably its low resolution when reconstructing
3D images of transparent specimens, especially along the optical axis. This limitation
is easily understood when remembering that only one illumination direction is used
in holographic microscopy, translating into limited quantity of recorded information,
so poor 3D image reconstructions [67]. Such a configuration indeed does not fulfill
the classical requirement of transmission microscopy that a high numerical aperture
condenser is to be used in conjunction with a high numerical aperture objective,
satisfying Köhler illumination, in order to obtain high-resolution image. Indeed, it
is not possible with coherent light to illuminate the sample under large numerical
aperture, as in that case a focal spot would be produced, resulting from the interference
of the various illuminations. So, DHM, strictly speaking, is in fact not an efficient
3D imaging technique for imaging transparent samples in transmission mode, and is
sometimes referred to as being a 2.5D imaging method, i.e., imaging a 2D surface
wrapped in a 3D volume.
In order to compensate for that limited resolution of DHM, phase tomography
has been proposed. Rotating the sample, one can record a set of 2D phase maps from
the sample observed under different angles, which are then converted into 3D index
of refraction maps, using classical backprojection of the data (Radon transform).
The observed sample itself can be easily rotated in the case of an optical fiber [68],
while free-standing samples like pollens, diatoms, amoeba, can be embedded within
a microcapillary, which is then rotated to perform phase tomography [69, 70].
However, this simplified approach neglects the effects of diffraction. So while
perfectly adapted at macroscopic scale, when characteristic sample features approach
wavelength dimension, as in microscopy, the image reconstruction quality is more
and more limited, especially for high-contrast samples. This limits the quality of
the reconstructed data in terms of quantitative index of refraction measurements, as
89
Demodulation is easily performed in Fourier space, in which the different orders of
diffraction are spatially separated. This technique has the advantage, compared to
phase-shifting holography, that a single hologram is enough to extract amplitude and
phase of the diffracted beam, but at the price of a reduced field of view, because of
the demodulation constraints in Fourier space.
Note that combined approached can also be used [52]. For more details about
digital holographic data acquisition and reconstruction, the interested reader will
find extensive description in [53–58].
Indeed, DHM is a very powerful technique of investigation when considering
integral measurements, by projections of the accumulated phase changes when light
travels through the observed samples [59], rapid phase variations, for example, corresponding to samples fluctuations [60, 61], very weak sample phase changes, thanks
to its high sensitivity, which can be linked to other physical quantities, for example,
allowing for optical patch clamp measurements [62]. It is also able to acquire long
time-lapse, without inducing photodamages to the sample, thanks to the low level
of light necessary to acquire interferograms (compared to confocal fluorescence),
which reveals useful for cell cycle, cell division, cell migration, drug influence or
cell apoptosis studies, cell detection and identification [63–66].
However, its main drawback is probably its low resolution when reconstructing
3D images of transparent specimens, especially along the optical axis. This limitation
is easily understood when remembering that only one illumination direction is used
in holographic microscopy, translating into limited quantity of recorded information,
so poor 3D image reconstructions [67]. Such a configuration indeed does not fulfill
the classical requirement of transmission microscopy that a high numerical aperture
condenser is to be used in conjunction with a high numerical aperture objective,
satisfying Köhler illumination, in order to obtain high-resolution image. Indeed, it
is not possible with coherent light to illuminate the sample under large numerical
aperture, as in that case a focal spot would be produced, resulting from the interference
of the various illuminations. So, DHM, strictly speaking, is in fact not an efficient
3D imaging technique for imaging transparent samples in transmission mode, and is
sometimes referred to as being a 2.5D imaging method, i.e., imaging a 2D surface
wrapped in a 3D volume.
In order to compensate for that limited resolution of DHM, phase tomography
has been proposed. Rotating the sample, one can record a set of 2D phase maps from
the sample observed under different angles, which are then converted into 3D index
of refraction maps, using classical backprojection of the data (Radon transform).
The observed sample itself can be easily rotated in the case of an optical fiber [68],
while free-standing samples like pollens, diatoms, amoeba, can be embedded within
a microcapillary, which is then rotated to perform phase tomography [69, 70].
However, this simplified approach neglects the effects of diffraction. So while
perfectly adapted at macroscopic scale, when characteristic sample features approach
wavelength dimension, as in microscopy, the image reconstruction quality is more
and more limited, especially for high-contrast samples. This limits the quality of
the reconstructed data in terms of quantitative index of refraction measurements, as