Biomedical Engineering Reference
In-Depth Information
provide a direct read-out of quantitative collagen structure. Future studies may incorporate both the
pol-scope and the changes mentioned in SHG collagen signal collection, providing more rigorous com-
parison between the techniques.
Collagen and its structure have been the focus of much research involving epithelial cell-based cancers
using SHG. However to date, no study compared the picrosirius red to SHG. Fundamentally, picrosirius
red has inferior resolution than SHG due to its dependence on the picric acid molecules binding to the col-
lagen fibers as well as the light source resolution limitations. Also picrosirius red requires fixation, limit-
ing it to one optical image plane, whereas SHG signal is innate to collagen and permits optical sectioning,
thus providing a 3D reconstruction of the microenvironment. Besides fundamental differences between
the techniques, it is quite difficult to obtain picrosirius red and SHG images from the same location within
the biopsy due to pathological sectioning used to compare the techniques. Despite these limitations, pic-
rosirius and other types of contrast imaging should be explored and compared with second-harmonic
imaging to potentially yield new information and perhaps provide more clinically accessible techniques.
Both picrosirius and SHG approaches are capable of visualizing wavy, straightened, and dense colla-
gen deposition in the tumor microenvironment. SHG seems to be more sensitive in less abundant regions
of collagen and displays a higher sensitivity to changes in waviness than picrosirius red. However, picro-
sirius red does show changes within collagen structures well enough to have potential clinical use since
it is much simpler, less expensive, and less disruptive to the current workflow in the clinic than SHG.
16.3 Quantifying SHG Data
16.3.1 the need for improved computational Methods
Collagens are the most abundant proteins in mammalian tissues and the major protein constituents of
the ECM, which principally maintains shape and structural integrity for cells and tissues, and plays an
important role in wound healing, tissue repair, and morphogenesis. Advanced imaging techniques have
increased our ability to study this complex relationship [29−31,36,45] and have led to the discovery of
some important phenomena [16,17,28,46,47]. Despite the sophisticated imaging techniques available
to examine the relationship between changes in collagen I and disease state—dermal wounds [48−50],
breast cancer [16,17,28,51,52], ovarian cancer [37], prostate cancer [53], asthma [54,55]—there is still
a dearth of robust computational methods for characterizing these changes. The development of new
computational analysis techniques will assist researchers in fully exploiting the potential of informa-
tion-rich imaging data. The techniques described below have the common goals of improving image
data analysis for biologists and developing techniques that will aid researchers in the exploration of
stromal collagen I as a biomarker for disease.
16.3.2 Digital Signal Processing for collagen Analysis
The images resulting from these microscopy techniques can be quantitatively analyzed using methods
based on digital signal processing. A signal can be defined as a description of the way that the behavior
of one process or parameter depends on another [47]. This relationship is typically represented as a
mathematical function of one or more independent variables [56]. A one-dimensional signal, such as
an audio signal, consists of an independent variable, which commonly represents time, distance, and
so on, and a dependent variable, which is a function of the independent variable (Figure 16.7). If the
independent variable occurs over a range of continuous values, then the signal is said to be a continuous
signal. If the independent variable occurs over a range of discrete values, then the signal is a discrete or
digital signal. For a digital signal, the independent variable describes the sampling of the process and
the dependent variable is the value of each sample [47].
The independent variable can also be said to define the domain of a signal. A signal for which the inde-
pendent variable represents time is a time-domain signal. Likewise, a signal for which the independent
