Biomedical Engineering Reference
In-Depth Information
Fig. 4.14
(
a1
-
a3
) Projection images obtained by reconstructing their corresponding SR holo-
grams are shown for three different illumination angles. These projection images are registered
with respect to the bead at the center of the FOV to demonstrate how the projection images change
as a function of angle for the same volume of the sample
common moveable bridge (arc-shaped plastic piece shown in Fig.
4.13
). Low-cost
magnets are attached on both ends of this bridge for electromagnetic actuation with
low power consumption. By applying different voltages across the coils (which
are mounted facing the magnets inside the housing), the movable bridge can be
translated, leading to simultaneous shifts of all the fibers along both the x and y
directions. Once fibers are shifted, a new set of projection images is obtained for all
the angles. The exact amount of displacement for these fiber-ends does not need to
be repeatable or accurately controlled as the hologram shifts are digitally estimated
with no prior knowledge required as explained in Sect.
4.5
. By recording 10-15
projection holograms, one SR hologram for each illumination angle is generated
(see Fig.
4.14
), which are used to digitally compute tomograms [
20
,
22
]. A set of 24
images can be acquired in
6 s at 4 frames/s, which can be significantly sped up
using a sensor with higher frame rate (e.g., >15-20 fps).
To demonstrate tomographic imaging using the device shown in see Fig.
4.13
,we
imaged microbeads of 5m diameter (refractive index
1:68) distributed within
achamber(
1:52). As
shown by the slice images in Fig.
4.15
(b1-b5), randomly distributed beads in the
chamber are successfully imaged in their corresponding depths as validated by
conventional microscope (40
50m thick) filled with optical gel (refractive index
, 0.65-NA) images. The inset in Fig.
4.15
,enclosed
with the dashed rectangle, further demonstrates the depth-sectioning performance,
where two axially overlapping microbeads are discerned. In a separate experiment
shown in Fig.
4.12
c3, the FWHM along the axial line-profile for a 2m bead
tomogram was measured as 7:8m[
20
]. These results suggest an axial resolution
of <7m, that is, >13
improvement over what is achieved using a single LR
hologram. Computing the tomograms for the region-of-interest (with
50m
thickness) shown in Fig.
4.15
takes <1 min using a graphical processing unit (GPU,
NVidia
Geforce GTX 480).
To investigate the performance of our field-portable lensfree tomographic micro-
scope for potential biomedical imaging applications, we conducted experiments
with a
Hymenolepis nana (H. nana)
egg, which is an infectious parasitic flatworm
having an approximately spherical shape. As demonstrated in Fig.
4.16
, computed
