Page 27 - Vibrational Spectroscopic Imaging for Biomedical Applications

Towar d Automated Br east Histopathology 5

28
replaced by rapid scanning techniques to achieve higher scanning
29
efficiency. Briefly, the process of data acquisition involves the inter-
ferometric encoding of a broadband blackbody emitter at high tem-
perature as a source. The output is guided to a microscope equipped
with all-reflecting optics. Since glass absorbs strongly, lenses are
reflective and specially coated mirrors are employed to permit collec-
tion of the wide mid-IR spectral region. The focusing optics utilizes
Cassegranian-type elements and typically condense the beam by a
factor of ~10 to 15. Detection is accomplished by liquid nitrogen
cooled mercury-cadmium-telluride (MCT) array detectors. Visible
images of a specimen are collected from the same field of view using
a white light source and a parfocal and collinear optical path.

1.1.2 FT-IR Spectroscopic Characterization
of Cells and Tissues
Efforts to analyze tissue with IR spectroscopy began nearly 60 years
ago with the first published diseased and normal breast, bladder, and
22
blood spectra. Early work in applying FT-IR spectroscopy for histo-
pathology recognition involved examining single spectra from large
tissue sections, DNA extracts and cell cultures. Although this work
was groundbreaking, the field did not immediately take off due to
the significant amount of tissue needed to acquire the spectra, primi-
tive instrument sensitivity, and difficulty in reproducing data. This
made meaningful spectral interpretation nearly impossible and the
work was not pursued for nearly 40 years. In the late 1980s, this line
of work was revived and led to notable publications on spectral
30
abnormalities in colon cancer. However, the low sample numbers,
uncertain tissue heterogeneity, and lack of reproducibility of simple
measures used to discriminate benign from malignant samples cast
doubts on the validity of these studies. Several other studies applied
31
FT-IR spectroscopy to discern premalignant tumor markers and
32
metastatic DNA features. While these works supported the concept
of monitoring cancer and its predisposition as well as understanding
of the importance of the microenvironment in tumor development,
they did not directly address the question of providing a diagnostic
measure that would appeal to clinicians.
Following the resurrection of interest in application of IR spec-
troscopy for disease diagnosis, breast cancer was one of the major foci
investigated using human tumor tissue samples, human tumor cell
23
lines, and xenografted human tumor cells. Most of the observed
spectral differences were attributed to tissue heterogeneity due to col-
lagen and fat content in connective tissues, emphasizing the need to
deconvolve the effects of tissue histology from that of pathology. The
relevance of collagen content in tumor samples due to the significant
overlap of collagen and DNA/RNA phosphate spectral features was
33
also recognized. This work emphasized the importance of considering