On-Chip Holographic Microscopy and its Application for Automated Semen Analysis - Biomedical Optical Phase Microscopy and Nanoscopy - page 158

the digital algorithms that this platform uses for automatic quantification of sperm density
and motility.

As the key component of this compact semen analysis platform, a self-contained on-chip
microscope (see Figure 8.1A ) is designed to record the holographic images of semen
samples over an FOV of B 24 mm 2 with an effective NA of B 0.1 0.2 without mechanical

scanning [10,11] . Inside this compact microscope assembly (weighs B 46 g and measures
B 4.2 cm 3 4.2 cm 3 5.8 cm, see Figure 8.1A ), a simple light-emitting diode is butt-coupled
to a 0.1 mm pinhole for casting partially coherent illumination over the semen sample,
which is B 4 cm away from the aperture plane (see Figure 8.1B ). Inside the same assembly,
a monochrome CMOS image sensor with an active area of 24 mm 2 records the lensless
holograms of the sperm and then transfers these raw images through a USB 2.0 connection
to a laptop computer, which could be replaced by a smartphone or personal digital assistant
(PDA) for better mobility. The semen samples are dispensed into disposable counting
chambers and then are loaded into this microscope with a sliding tray, which holds the
chamber at B 1 mm above the active area of the CMOS image sensor (see Figure 8.1A ).

For performing automated sperm analysis with this lensless microscope, 20 consecutive
lensless holographic frames are recorded for each semen sample at a frame rate of B 2
frames per second. The integration time of each frame is set to , 35 ms for minimizing the
motion blur of motile sperm. In order to separately characterize the immotile and the motile
sperm in the semen sample, two different processing approaches, that is, digital summation
and subtraction of these lensless holographic frames, were applied, respectively.

For identification and counting of immotile sperm, all the individual holographic frames are
first normalized and summed up digitally. This summation operation not only increases the
signal-to-noise ratio (SNR) of the immotile sperm' holographic patterns but also smears out
the patterns from the motile sperm. This critical step creates significant contrast
improvement on the faint signatures of the sperm' tails and enables automated identification
of the immotile sperm through their tail signatures (see Figure 8.2C and D ).

Next, this summation hologram is processed by the iterative holographic reconstruction
algorithm described in Section 8.3 to generate the microscopic amplitude and phase images
of the immotile sperm (see Figure 8.2B and C ). For automatic counting of immotile sperm,
distinct and bright elliptical heads in the reconstructed phase images are initially isolated
from the background as candidate patterns. The isolation process involves a thresholding
operation, where pixels above a certain intensity value were grouped together, and a
morphological screening process, where several properties of the connected regions such as
pixel area, orientation, and circularity are analyzed [12] . After this initial head isolation
step, the tail of each sperm also needs to be identified with its length, location, and
orientation. Since the presence of a healthy tail is necessary for a viable sperm count
according to the WHO laboratory manual [18] , the tail in the reconstructed phase image