Optical Imaging in Biology: Basics and Applications
657
matrix [79]. Since spectral data can be expressed as points in hyperspace, spectrally similar pixels can cluster together, and chemometric algorithms can be used
to distinguish such clusters for automated end-member discrimination [79]. Clustering algorithms can be used for data mining for supervised learning or indirect
knowledge discovery. Several clustering algorithms such as K-means clustering has
been attempted in biomedical spectroscopy which can classify clusters by minimizing Euclidean distances between points in hyperspace. Additionally, support
vector machines (SVMs) are also used as learning machines for the classification
of a spectral data set [80, 81]. A few classifiers used for imaging analysis is shown in
Fig. 9. SVMs and machine learning algorithms have been used for cancer imaging,
blood cell detection, and a host of other biomedical applications [79].
9 Conclusion and Future Outlook
Optical imaging is an invaluable tool in many fields of biology and has greatly
contributed to our understanding of the cellular processes. The basic properties
of light-matter interactions such as scattering, absorbance, polarization, coherence,
lifetime, and non-linear effects have been utilized heavily to supplement the everincreasing quest for a better understanding of the inner workings of cells and their
communication. As discussed, these methods are widely available, relatively inexpensive, and readily adaptable for live analysis to follow cellular pathways in living
cells with minimal disturbance. Imaging is also used to answer many basic and
clinical biological questions with the help of antibody-coupled fluorophores that
can diffuse through the cell membranes. With the integration of correlation and
lifetime imaging, biomedical research is moving slowly towards the imagery of
live specimens for real-time quantitative evaluation. The use of NIR fluorescence
imaging for a variety of ophthalmologic indications is one successful example of the
translation of optical imaging where fundus autofluorescence was used for noninvasively imaging the endogenous ocular fluorophore lipofuscin in the retinal pigment
epithelium [83]. Label-free imaging modalities focused on utilizing the inherent
light-tissue interaction properties constitutes another cornerstone of this field. In
contrast to a two-dimensional snapshot, these methods can provide information
about the biochemical composition as well. These methods have been utilized to
study various clinically relevant physiological processes as well as the quantitative
analysis of cellular dynamics. With the continuous development of instrumentation
and image processing algorithms, the gradual implementation of these technologies
in a point-of-care environment to provide real-time diagnostic information seems
feasible.
Although the field of optical imaging has gained tremendous momentum in the
last two decades, it will take some time for the scientific community to embrace these
methods fully and for widespread availability as commercial instruments. Several
technical criteria had to be met before an imaging technique could be used for biological applications. It involves the choice of an appropriate light source of sufficient
657
matrix [79]. Since spectral data can be expressed as points in hyperspace, spectrally similar pixels can cluster together, and chemometric algorithms can be used
to distinguish such clusters for automated end-member discrimination [79]. Clustering algorithms can be used for data mining for supervised learning or indirect
knowledge discovery. Several clustering algorithms such as K-means clustering has
been attempted in biomedical spectroscopy which can classify clusters by minimizing Euclidean distances between points in hyperspace. Additionally, support
vector machines (SVMs) are also used as learning machines for the classification
of a spectral data set [80, 81]. A few classifiers used for imaging analysis is shown in
Fig. 9. SVMs and machine learning algorithms have been used for cancer imaging,
blood cell detection, and a host of other biomedical applications [79].
9 Conclusion and Future Outlook
Optical imaging is an invaluable tool in many fields of biology and has greatly
contributed to our understanding of the cellular processes. The basic properties
of light-matter interactions such as scattering, absorbance, polarization, coherence,
lifetime, and non-linear effects have been utilized heavily to supplement the everincreasing quest for a better understanding of the inner workings of cells and their
communication. As discussed, these methods are widely available, relatively inexpensive, and readily adaptable for live analysis to follow cellular pathways in living
cells with minimal disturbance. Imaging is also used to answer many basic and
clinical biological questions with the help of antibody-coupled fluorophores that
can diffuse through the cell membranes. With the integration of correlation and
lifetime imaging, biomedical research is moving slowly towards the imagery of
live specimens for real-time quantitative evaluation. The use of NIR fluorescence
imaging for a variety of ophthalmologic indications is one successful example of the
translation of optical imaging where fundus autofluorescence was used for noninvasively imaging the endogenous ocular fluorophore lipofuscin in the retinal pigment
epithelium [83]. Label-free imaging modalities focused on utilizing the inherent
light-tissue interaction properties constitutes another cornerstone of this field. In
contrast to a two-dimensional snapshot, these methods can provide information
about the biochemical composition as well. These methods have been utilized to
study various clinically relevant physiological processes as well as the quantitative
analysis of cellular dynamics. With the continuous development of instrumentation
and image processing algorithms, the gradual implementation of these technologies
in a point-of-care environment to provide real-time diagnostic information seems
feasible.
Although the field of optical imaging has gained tremendous momentum in the
last two decades, it will take some time for the scientific community to embrace these
methods fully and for widespread availability as commercial instruments. Several
technical criteria had to be met before an imaging technique could be used for biological applications. It involves the choice of an appropriate light source of sufficient