448
K. malek et al.
this algorithm provides spectral features of normal cervical transformation zone,
associated with anatomical features observed in h&E staining such as the ectocervical and endocervical epithelium, the squamous junction, and red blood cells. the
squamous layer shows high concentration of glycogen, whereas the epithelium tissue exhibits the presence of glycoproteins. A good demarcation pattern of squamous
epithelium layers was also found for a patient diagnosed with low-grade squamous
intraepithelial lesions [100]. here, most of the differentiation was due to variations
in protein bands. Furthermore, the superficial layer exhibits a high glycogen content
in contrast with the parabasal layer, in which a reduction in the amide I/amide II
ratio is also observed. As expected, IR spectra of the connective tissue show strong
collagen bands and no contribution from glycogen. the unexpected low glycogen
content was found within the parabasal and superficial layer, which was proposed
as an indicator of parakeratosis. Below this layer, an identified cluster was associated with inflammation. Furthermore, this approach clearly detects abnormal areas
within a tissue section diagnosed as CINII/III. the spectrum representative of such
a metastasis highlights a significant decrease in amide II/amide I intensity ratio,
suggesting that the amide II band can be a possible marker of diseased tissue. however, one cannot rule out the possibility that the excellent demarcation of cancer
versus normal tissue in these sections was the result of light dispersion because of
the different refractive indices observed between the highly nuclei dense regions
of CIN III dysplasia compared to the normal tissue with pyknotic nuclei and large
amounts of cytoplasm. the extracted spectra from these maps did show a significant shift in the amide I region indicating evidence of dispersion. given that the best
demarcation was achieved using the amideI/amide II region and not the phosphodiester region (1200–1000 cm
−1
) and at that point in time Rmie-EmSC had not been
developed it is reasonable to assume that the demarcation was based on dispersion
and not differences in protein expression. other spectral features of this abnormal
tissue include an increase in intensity of both phosphodiester bands [ν as (Po 2
−
) and
ν s (Po 2
−
)]. Figure 15.7 shows h&E stained sections (overview and zoomed in area)
of a CIN III diagnosed cervical tissue clearly highlighting regions of cancerous
tissue in the darker purple colour. Below are the corresponding FtIR maps after
performing unsupervised hierarchical Cluster Analysis (uhCA) on the amide I
and amide II region (1700–1500 cm
−1
). the uhCA maps clearly delineate the major
histological and histopathological features. the brown is associated with the abnormal cells, light blue lymphocyte exudates, dark blue connective tissue, orange red
blood cells and the red are regions are areas of weak absorbance. As shown in the
bottom right panel this approach also resolves micro metastases and surrounding
lymphocytes. this work has clearly demonstrated that combination of FtIR imaging with unsupervised multivariate image analysis provides a “blind” detection of
biological variability within normal and diseased state of the cervical tissue [100]
albeit based on physico-chemical differences and not solely chemical differences.
Steller and co-workers performed a similar study on a sample from a patient
with a squamous cell carcinoma of the uterine cervix who underwent a radical hysterectomy [101]. 122 chemical images of the tissue were recorded using an FtIR
microspectrometer coupled with a 64 × 64 FPA detector. Spectral processing was
K. malek et al.
this algorithm provides spectral features of normal cervical transformation zone,
associated with anatomical features observed in h&E staining such as the ectocervical and endocervical epithelium, the squamous junction, and red blood cells. the
squamous layer shows high concentration of glycogen, whereas the epithelium tissue exhibits the presence of glycoproteins. A good demarcation pattern of squamous
epithelium layers was also found for a patient diagnosed with low-grade squamous
intraepithelial lesions [100]. here, most of the differentiation was due to variations
in protein bands. Furthermore, the superficial layer exhibits a high glycogen content
in contrast with the parabasal layer, in which a reduction in the amide I/amide II
ratio is also observed. As expected, IR spectra of the connective tissue show strong
collagen bands and no contribution from glycogen. the unexpected low glycogen
content was found within the parabasal and superficial layer, which was proposed
as an indicator of parakeratosis. Below this layer, an identified cluster was associated with inflammation. Furthermore, this approach clearly detects abnormal areas
within a tissue section diagnosed as CINII/III. the spectrum representative of such
a metastasis highlights a significant decrease in amide II/amide I intensity ratio,
suggesting that the amide II band can be a possible marker of diseased tissue. however, one cannot rule out the possibility that the excellent demarcation of cancer
versus normal tissue in these sections was the result of light dispersion because of
the different refractive indices observed between the highly nuclei dense regions
of CIN III dysplasia compared to the normal tissue with pyknotic nuclei and large
amounts of cytoplasm. the extracted spectra from these maps did show a significant shift in the amide I region indicating evidence of dispersion. given that the best
demarcation was achieved using the amideI/amide II region and not the phosphodiester region (1200–1000 cm
−1
) and at that point in time Rmie-EmSC had not been
developed it is reasonable to assume that the demarcation was based on dispersion
and not differences in protein expression. other spectral features of this abnormal
tissue include an increase in intensity of both phosphodiester bands [ν as (Po 2
−
) and
ν s (Po 2
−
)]. Figure 15.7 shows h&E stained sections (overview and zoomed in area)
of a CIN III diagnosed cervical tissue clearly highlighting regions of cancerous
tissue in the darker purple colour. Below are the corresponding FtIR maps after
performing unsupervised hierarchical Cluster Analysis (uhCA) on the amide I
and amide II region (1700–1500 cm
−1
). the uhCA maps clearly delineate the major
histological and histopathological features. the brown is associated with the abnormal cells, light blue lymphocyte exudates, dark blue connective tissue, orange red
blood cells and the red are regions are areas of weak absorbance. As shown in the
bottom right panel this approach also resolves micro metastases and surrounding
lymphocytes. this work has clearly demonstrated that combination of FtIR imaging with unsupervised multivariate image analysis provides a “blind” detection of
biological variability within normal and diseased state of the cervical tissue [100]
albeit based on physico-chemical differences and not solely chemical differences.
Steller and co-workers performed a similar study on a sample from a patient
with a squamous cell carcinoma of the uterine cervix who underwent a radical hysterectomy [101]. 122 chemical images of the tissue were recorded using an FtIR
microspectrometer coupled with a 64 × 64 FPA detector. Spectral processing was