450
K. malek et al.
the validity of FtIR microspectroscopy for classifying biopsies with symptoms
of cervical intraepithelial neoplasia in different grading (CIN I-III) has been evaluated by mordechai and co-workers by using probabilistic neural networks (PNNs)
and histopathology [102]. the results show the high level of accuracy in discrimination of the grading of neoplasia (CINI, CINII, and CINIII) comparable to a histological method. the false positive identification of the normal as CINI, CINII,
and CINIII patients was 9.04, 0.01, and 0.01 %, respectively. the small errors were
encouraging for further advanced studies on spectroscopic diagnostics [102]. the
same research group has reported that spectral features of a control group and a
group of patients diagnosed with cervical cancer result from the decrease in the
glucose/phosphate ratio (by 23–49 %) and the increase in the RNA/dNA ratio (by
38–150 %) in carcinogenic compared with normal tissue [103]. In this case, probabilistic neural networks and statistical tests predicted the presence of cancer with a
high degree of accuracy (> 95 %).
In another experiment, Bogomolny et al. [104] have studied the application
of FtIR microspectroscopy for monitoring malignancy progression in murine fibroblast (NIh/3t3) and mouse embryonic (mEF) cells infected with murine sarcoma virus (muSv). human cervical tissues at different stages of neoplasia were
used for testing. they proposed an array of four biomarkers, which discriminated
different premalignant stages with a classification accuracy of 89.5–97.4 %. the
identified biomarkers are associated with alterations in the lipids/proteins [A 2958 /
(A 2852 + A 2923 )], RNA/dNA (A 1121 /A 1015 ), protein/carbohydrates (A 1171 /A 1152 ), and
RNA/carbohydrates [(A 1082 -A 1056 )/(A 1028 )] ratios. this enabled further divisions
of cancerogenous transformation of the cell lines into three premalignant stages
according to the t-value of the discriminate classification score (dCS). only the
last stage of the disease progression exhibited morphological signs of malignancy,
showing de facto an advantage of FtIR microscopy over standard techniques in
the cell examination. In general, the analysis of biomarkers derived from spectra
of cells and different neoplastic stages of cervical tissue (CINI/II/III) indicates the
decrease in lipid, dNA, and carbohydrates content during the cancer progression,
whereas the absorptions of proteins and RNA have a tendency to increase gradually.
these results were found in the agreement with other studies on detection of malignancy from FtIR spectra. moreover, this approach showed relatively high accuracy
( ca. 90 %) and specificity ( ca. 92 %) in classification of biopsies of cervical tissue
from 37 patients diagnosed as invasive cervical carcinoma [104].
An attempt towards discrimination of a spectral phenotype hPv-infected cervical tissue has been undertaken by using AtR FtIR spectroscopy [105]. PCA-LdA
analysis did not categorise low-grade from high-grade squamous intraepithelial lesions associated with hPv infection types. however, in a group of low-grade cytologically diagnosed with high-risk oncogenic hPv16/18, the multivariate analysis discriminated women based on their age into the two classes (20–29 years vs.
30–39 years). PCA loading plot exhibited the most prominent spectral differences
between the both groups corresponding to vibrations of proteins (amide II and III
bands, a band at 1404 cm
−1
), phosphorylated proteins (972 cm
−1
), and carbohydrates
(1153 cm
−1
). Spectral features specific for 4
th
decade women were those associated
K. malek et al.
the validity of FtIR microspectroscopy for classifying biopsies with symptoms
of cervical intraepithelial neoplasia in different grading (CIN I-III) has been evaluated by mordechai and co-workers by using probabilistic neural networks (PNNs)
and histopathology [102]. the results show the high level of accuracy in discrimination of the grading of neoplasia (CINI, CINII, and CINIII) comparable to a histological method. the false positive identification of the normal as CINI, CINII,
and CINIII patients was 9.04, 0.01, and 0.01 %, respectively. the small errors were
encouraging for further advanced studies on spectroscopic diagnostics [102]. the
same research group has reported that spectral features of a control group and a
group of patients diagnosed with cervical cancer result from the decrease in the
glucose/phosphate ratio (by 23–49 %) and the increase in the RNA/dNA ratio (by
38–150 %) in carcinogenic compared with normal tissue [103]. In this case, probabilistic neural networks and statistical tests predicted the presence of cancer with a
high degree of accuracy (> 95 %).
In another experiment, Bogomolny et al. [104] have studied the application
of FtIR microspectroscopy for monitoring malignancy progression in murine fibroblast (NIh/3t3) and mouse embryonic (mEF) cells infected with murine sarcoma virus (muSv). human cervical tissues at different stages of neoplasia were
used for testing. they proposed an array of four biomarkers, which discriminated
different premalignant stages with a classification accuracy of 89.5–97.4 %. the
identified biomarkers are associated with alterations in the lipids/proteins [A 2958 /
(A 2852 + A 2923 )], RNA/dNA (A 1121 /A 1015 ), protein/carbohydrates (A 1171 /A 1152 ), and
RNA/carbohydrates [(A 1082 -A 1056 )/(A 1028 )] ratios. this enabled further divisions
of cancerogenous transformation of the cell lines into three premalignant stages
according to the t-value of the discriminate classification score (dCS). only the
last stage of the disease progression exhibited morphological signs of malignancy,
showing de facto an advantage of FtIR microscopy over standard techniques in
the cell examination. In general, the analysis of biomarkers derived from spectra
of cells and different neoplastic stages of cervical tissue (CINI/II/III) indicates the
decrease in lipid, dNA, and carbohydrates content during the cancer progression,
whereas the absorptions of proteins and RNA have a tendency to increase gradually.
these results were found in the agreement with other studies on detection of malignancy from FtIR spectra. moreover, this approach showed relatively high accuracy
( ca. 90 %) and specificity ( ca. 92 %) in classification of biopsies of cervical tissue
from 37 patients diagnosed as invasive cervical carcinoma [104].
An attempt towards discrimination of a spectral phenotype hPv-infected cervical tissue has been undertaken by using AtR FtIR spectroscopy [105]. PCA-LdA
analysis did not categorise low-grade from high-grade squamous intraepithelial lesions associated with hPv infection types. however, in a group of low-grade cytologically diagnosed with high-risk oncogenic hPv16/18, the multivariate analysis discriminated women based on their age into the two classes (20–29 years vs.
30–39 years). PCA loading plot exhibited the most prominent spectral differences
between the both groups corresponding to vibrations of proteins (amide II and III
bands, a band at 1404 cm
−1
), phosphorylated proteins (972 cm
−1
), and carbohydrates
(1153 cm
−1
). Spectral features specific for 4
th
decade women were those associated