Biomedical Engineering Reference
In-Depth Information
Normal and diseased breast tissues could be differentiated from the contribution
of carotenoids and lipids. A small contribution from a heme-type signal was
detected in some samples of clinically abnormal yet histopathologically benign
breast tissue, while a much stronger heme-type signal was detected in most of the
breast cancers. Raman spectra of diseased breast tissue (benign and malignant) also
showed diminished to absent contributions from lipids and reduced contributions
from carotenoids. Haka et al. [
91
] did a thorough
ex vivo
study, in which 130
spectra from 58 patients were obtained. Model fitting was performed by using a
linear combination of the contribution of each base spectrum with a nonnegativity
constraint. Statistical analysis showed a sensitivity of 94% and specificity of 96%.
Very recently, the authors also designed a ball-lens-based fiber probe for
in vivo
breast cancer diagnosis [
92
]. Statistical analysis showed 100% sensitivity and 100%
specificity. However, this result was based on very limited samples (21 normal,
8 fibrocystic, and 1 cancer from 9 patients).
1.4.7
Cervical Cancer Diagnosis
Cervical cancer is the second most common malignancy among women world-
wide. In recent years, diffuse reflectance, fluorescence, and Raman spectroscopy
have been used for cervical cancer and precancer screening and diagnosis [
3
,
4
].
Mahadevan-Jansen et al. first studied the Raman properties of
ex vivo
human
cervical tissue using NIR Raman spectroscopy [
93
]. Thirty-six biopsies were
obtained from 18 patients. The authors found that precancer tissues could be
well differentiated from normal tissue with an average sensitivity of 88% and
specificity of 92% and high-grade from low-grade cervical cancers with a sensitivity
and specificity of 100%. Utzinger et al. studied the NIR Raman spectroscopy of
cervical cancer
in vivo
using an optical fiber-based Raman probe [
44
,
94
]. Twenty-
four
in vivo
spectra were obtained from 13 patients. Spectra measured
in vivo
resemble those measured
in vitro
. The main Raman peaks for cervical tissue located
in the vicinity of 1,070, 1,180, 1,195, 1,210, 1,245, 1,330, 1,400, 1,454, 1,505,
1,555, 1,656, and 1;760 cm
1
. The ratio of intensities at 1;454-1;656 cm
1
and
1;330-1;454 cm
1
could be used to differentiate squamous dysplasia from other
tissue types. Recently, Mo et al. designed a ball-lens-based optical fiber probe
for
in vivo
cervical cancer diagnosis [
45
]. They measured the Raman spectra of
cervical tissue
in vivo
in the high wavenumber region. A total of 92
in vivo
Raman
spectra (46 normal, 46 dysplasia) were obtained from 46 patients with Pap smear
abnormalities of the cervix. Significant difference was observed in the Raman
intensities of the prominent Raman bands at 2,850 and 2;885 cm
1
(CH
2
stretching
of lipids), 2;940 cm
1
(CH
3
stretching of proteins), and the broad Raman band of
water (peaking at 3;400 cm
1
in the 3;100-3; 700
cm
1
range) between normal and
dysplasia cervical tissue. The statistical analysis based on PCA and LDA together
with the leave-one-out cross-validation method showed a sensitivity of 93.5% and
specificity of 97.8% for dysplasia tissue identification.
Search WWH ::
Custom Search