Page 258 - Vibrational Spectroscopic Imaging for Biomedical Applications

234 Cha pte r Ei g h t

characterized by spectral contributions due to unsaturated fatty acids
−1
−1
(1065, 1131, 1298, 1669 cm ), the ester group (1739 cm ) and the choles-
−1
terol moiety (701 cm ). Further Raman bands which distinguished
9
cholesterol ester and cholesterol can be found at 428, 538, and
−1
614 cm . Interestingly, the cholesterol ester content in normal brain
tissue is very low and significantly increases (up to 100 fold) in brain
tumors. Therefore, the detection of cholesterol ester can be consid-
ered as a molecular marker of brain tumors. This observation has
recently been confirmed in mass spectra of lipid extracts of brain
18
tumors. Raman spectra with higher and lower spectral contribu-
tions of hemoglobin were found to be the two main clusters in GBM.
Hemoglobin in dried tissue sections contributed to Raman bands at
−1
661, 751, 1004, 1124, 1258, 1346, and 1603 cm . These bands signifi-
cantly deviate from hemoglobin in fresh, nondried brain tumors (see
Fig. 8.5) and in red blood cells (see Fig. 8.9 in Sec. 8.3.2). The Raman
signature of hemoglobin in Fig. 8.4 (vine spectrum) is typical for the
heme group in an aggregated and denatured state. Spectral contribu-
tions of hemoglobin in GBM are consistent with high blood perfusion
rate which is a typical property of malignant brain tumors.
Nondried Brain Tissue
The investigations of thin, dried tissue sections have been transferred
17
to ex vivo studies of nondried brain tissue. These samples do not
show artifacts due to drying such as crystallization of hydrophobic
constituents, denaturation of biomolecule structures or shrinkage
after dehydration. Analogous to the investigation of mouse brains,
the specimens were covered by a window to prevent them from dry-
ing. Instead of a fiber-optic probe, the sample was placed under a
microscope with a 10×/NA 0.25 objective. The lower magnification
objective focuses the laser to a spot of approximately 50 μm diameter.
2
Raman images of 2 × 2 mm areas were collected with a step size of
100 μm. Cluster analyses were performed in the spectral interval
−1
1200 to 1800 cm because it contains spectral contributions of the
−1
−1
main constituents water (1635 cm ), proteins (1660 cm ) including
−1
−1
−1
collagen (1247 cm ), hemoglobin (1566 cm ), and lipids (1440 cm )
−1
including cholesterol (1670 cm ). A Raman image of GBM and the
spectra of the main clusters are shown in Fig. 8.5. Both spectra are
−1
dominated by protein bands near 1005, 1447, and 1662 cm . The
spectra differ by spectral contributions of hemoglobin at 755, 1004,
−1
1214, 1547, 1564, 1607, and 1621 cm . The intense hemoglobin bands
are attributed to a resonance effect of the chromophor heme which is
described in more detail in the context of red blood cells (Sec. 8.3.2).
Chemical analysis of brain tumors and Raman spectra of brain lipid
18
extracts confirmed that GBM are characterized by increased water
content and decreased lipid content. More subtle differences are
increased cholesterol ester and phosphatidylcholine contents. These
differences constitute the basis for a Raman-based classification