Page 197 - Vibrational Spectroscopic Imaging for Biomedical Applications

W idefield Raman Imaging of Cells and T issues 173

67
types of kidney tissue. The study evaluated the application of wide-
field Raman imaging to the differentiation of renal oncocytoma (OC)
from chromophobe renal cell carcinoma (ChRCC), which can often have
overlapping histological features such that a pathologist is not readily
able to distinguish the two types of diseases. A pathologist chose 10
cases of OC and 10 cases of ChRCC. Typical OC and ChRCC regions
within each tissue were highlighted by the pathologist. Widefield
Raman images were collected from these regions in unstained tissue
sections. PCA was employed to reduce the large widefield Raman
image datasets to smaller, more comprehensible results. Each point on
the scatter plot represents a spectrum derived from the widefield Raman
image: the red squares represent spectra obtained from the ChRCC and
the turquoise circles represent the spectra obtained from OC. The spec-
tral regions used in this evaluation were within the fingerprint region
-1
(1633 to 1850 cm ). In the plot, the axes are PC5 and PC3. This evalua-
tion shows a clear distinction between ChRCC and OC data, establish-
ing that widefield Raman imaging is an objective method that may
provide information to assist the pathologist in a diagnosis.
Similar to the kidney example, distinguishing between invasive
ductal carcinoma (IDC) and invasive lobular carcinoma (ILC) in
breast tissue is sometimes not possible without the benefit of further
staining. Figure 6.6 shows the scatter plot of a PC score of spectra
68
obtained from widefield Raman images of IDC and ILC. There is a
slight distinction between the two classes, although the separation
is not as good as the kidney tissue. ILC and IDC may not separate
well because a subset of ILC cells closely resembles IDC cells in
69
their transcription pattern. These preliminary results indicate that
widefield Raman imaging can indicate similarities in different tissue
types as well as differences. From the kidney and breast examples,
it is apparent that a tool such as widefield Raman imaging can
potentially assist pathologists in difficult diagnoses.

6.5.2 Mahalanobis and Euclidean Distance
PCA provides a basis on which models can be created to compare
different tissues. As described in the previous section, PCA was useful
in highlighting the spectral differences in tissue that are not easily
identified visually by a pathologist. The spectral differences in the
tissue are due to the collection of molecules that are present within
the cell or tissue; and that collection of biomolecules will be dictated
by the metabolic state of the cell or tissue. Therefore, widefield Raman
imaging and PCA should be useful in showing the spectral differences
between tissues in different metabolic states. Different metabolic states
will exist when tissue is inflamed, infected, cancerous, or in some
other diseased state. Models can be created from spectra derived from
widefield Raman images of biological tissue using PCA to differentiate
objectively the different metabolic state of cells or tissue.