Biomedical Engineering Reference
In-Depth Information
6.3.3.2
Boundary Extraction
After the noise reduction, the cell boundary was detected as a thick line (>1 pixel
width) by applying a 3 × 3 Sobel fi lter (Fig.
6.3d
) and Otsu thresholding (Fig.
6.3e
).
The 8-connected contour pixels were extracted from the thick line by skeletonization
(Fig.
6.3f
). As seen in Fig.
6.3g
, the image processing steps accurately approximated
the cell boundaries. The boundary evolution shown in Fig.
6.3h
suggests that
protrusions in the order of 1
m can be detected from the extracted cell boundary
with a time evolution in the order of seconds.
ʼ
6.3.4
Detection of Time Evolution of Cell Boundary
Quantitative characterization of cell morphological dynamics is a typical technical
requirement for a number of researchers. For this aim, a variety of techniques for
quantitation of time evolution of cell boundary have been developed. One major
consideration in choosing the method is the association between segments on the
boundary in a timelapse image sequence.
Using circular mapping (Fig.
6.4
) of the cell shape in a 2D Cartesian system, we
can describe the time evolution of an entire cell boundary. For cells with circular
morphology, in particular, it is an easy-to-use and a powerful approach. The method
of circular mapping successfully characterize discontinuous cell spreading that
occurs in an anisotropic manner through stochastic, transient extension periods,
named STEPs (Dubin-Thaler et al.
2004
), and also the ordered patterns in a spontaneous
cell migration (Maeda et al.
2008
). It can also be applied to collective cell protrusive
behavior after the boundary release controlled by a substrate with switchable
adhesive patterns (Rolli et al.
2012
). We can adopt the polar circular mapping-based
approach to characterize cell morphological dynamics of full-moon-shaped kerato-
cytes because these cells have a nearly circular morphology.
q
(°)
90
a
b
r
( ) = [
x
(
s
),
y
(
s
)]
c
12
0
6
0
s
20
30
s
10
0
0
q
r
(
)
y
210
330
x
240
300
270
Fig. 6.4
Circular mapping of cell shape (Miyoshi and Adachi
2012
). (
a
) Phase contrast raw image
of a keratocyte. In the analysis, (
b
) cell outline curve
r
= [
x
(
s
),
y
(
s
)] consisting of 8-connected
contour pixels is expressed in (
c
) polar coordinates,
r
(
ʸ
) (Adapted with permission from The Royal
Society of Chemistry: [Integrative Biology], copyright (2012))