17 Vibrational Microspectroscopy for Analysis of Atherosclerotic Arteries
517
reference spectra: Independent Component Analysis (ICE) or vertex Component
Analysis (vCA). Both of which use statistical concepts to derive abstract spectra
with abstract concentration profiles. this is just a short description of the very idea
of the several approaches available to a spectroscopist trying to analyze his data.
Interested readers are suggested to search for more specific literature.
17.3.2 Spectral Features of Atherosclerosis
17.3.2.1 Raman Spectroscopy
Classification of types of Atherosclerotic Plaque
Raman microspectroscopy studies on atherosclerosis can be divided into separate
topics depending on the type of plaque. the main goal of the majority of previous research was the identification of the atherosclerotic plaque and analysis of its
biochemical composition [82]. Some of this focused on the analysis of atheroma
core with the cholesterol and cholesteryl esters, as well as the analysis of the calcification at the outer base of advanced lesions. there are number of components of
atherosclerosis that are easily detected by Raman spectroscopy and could be used
in high-sensitivity atherosclerosis diagnosis with the use of this method. these include: collagen, cholesterol crystals, adipocytes and calcium crystals [2].
In a large number of Raman studies PCA was used to classify coronary artery tissue as either non-atherosclerotic, atherosclerotic, calcified or non-calcified plaque.
this method, however, can deliver just the basic information about the plaque and
does not provide the information about chemical composition of the lesion [82–84].
In most plaque samples, it was possible to correctly adjust them to the aforementioned types of plaque, when the model of Logistic regression was applied to the
first and third principle components [84]. the discriminant analysis based on mahalanobis distance, which uses distance to associate a characteristic from a sample
to a specific class, is suggested to be used instead of Euclidean distance [2, 82]. the
main advantage of mahalanobis distance is that it takes into account not only the
relative distance from the sample to the mean of the group, but also the covariance
of the data. this results in the statistical measurement of how well the spectral parameters of a sample match the original one [2]. Independent component analysis
(ICA) with the use of mahalanobis distance gave similar results to PCA and also
classifies the tissue to the general plaque groups [85].
to the group of more advanced diagnostic algorithms, which are utilized to classify plaques in coronary arteries according to their detailed, chemical composition,
we can include the second linear least squares-minimalization (LSm) model [83]. It
is interesting that in this method, the biochemical analysis is performed by modeling the Raman tissue spectra using a simple linear combination of the basis Raman
spectra of the major components presented in the tissue. Such components include
collagen type III, elastin, actin, myosin, tropomyosin, cholesterol monohydrate,
cholesteryl linoleate, phosphatidyl choline, triolin, β-carotene, calcium carbonate
517
reference spectra: Independent Component Analysis (ICE) or vertex Component
Analysis (vCA). Both of which use statistical concepts to derive abstract spectra
with abstract concentration profiles. this is just a short description of the very idea
of the several approaches available to a spectroscopist trying to analyze his data.
Interested readers are suggested to search for more specific literature.
17.3.2 Spectral Features of Atherosclerosis
17.3.2.1 Raman Spectroscopy
Classification of types of Atherosclerotic Plaque
Raman microspectroscopy studies on atherosclerosis can be divided into separate
topics depending on the type of plaque. the main goal of the majority of previous research was the identification of the atherosclerotic plaque and analysis of its
biochemical composition [82]. Some of this focused on the analysis of atheroma
core with the cholesterol and cholesteryl esters, as well as the analysis of the calcification at the outer base of advanced lesions. there are number of components of
atherosclerosis that are easily detected by Raman spectroscopy and could be used
in high-sensitivity atherosclerosis diagnosis with the use of this method. these include: collagen, cholesterol crystals, adipocytes and calcium crystals [2].
In a large number of Raman studies PCA was used to classify coronary artery tissue as either non-atherosclerotic, atherosclerotic, calcified or non-calcified plaque.
this method, however, can deliver just the basic information about the plaque and
does not provide the information about chemical composition of the lesion [82–84].
In most plaque samples, it was possible to correctly adjust them to the aforementioned types of plaque, when the model of Logistic regression was applied to the
first and third principle components [84]. the discriminant analysis based on mahalanobis distance, which uses distance to associate a characteristic from a sample
to a specific class, is suggested to be used instead of Euclidean distance [2, 82]. the
main advantage of mahalanobis distance is that it takes into account not only the
relative distance from the sample to the mean of the group, but also the covariance
of the data. this results in the statistical measurement of how well the spectral parameters of a sample match the original one [2]. Independent component analysis
(ICA) with the use of mahalanobis distance gave similar results to PCA and also
classifies the tissue to the general plaque groups [85].
to the group of more advanced diagnostic algorithms, which are utilized to classify plaques in coronary arteries according to their detailed, chemical composition,
we can include the second linear least squares-minimalization (LSm) model [83]. It
is interesting that in this method, the biochemical analysis is performed by modeling the Raman tissue spectra using a simple linear combination of the basis Raman
spectra of the major components presented in the tissue. Such components include
collagen type III, elastin, actin, myosin, tropomyosin, cholesterol monohydrate,
cholesteryl linoleate, phosphatidyl choline, triolin, β-carotene, calcium carbonate