Detecting Breast Cancer Using Raman Spectroscopy



Detecting Breast Cancer Using Raman Spectroscopy

The Project

The overall goal of this research project is to develop near-infrared (NIR) Raman spectroscopy as a histochemical/morphological tool for detection and diagnosis of breast cancer. We strive to characterize the Raman spectra of normal and diseased tissue types, correlate spectroscopic features with biochemistry and tissue morphology, establish diagnostic decision schemes, and develop instrumentation needed for rapid and accurate data collection and analysis both in vitro and in vivo.

Background

Breast cancer is the most common malignant tumor among women in the western world. In the US, approximately 180,000 new cases are diagnosed each year and 44,000 women die from the disease. Mammography is the most common technique for detecting non-palpable, highly curable breast cancer. In cases where the tissue is particularly dense, ultrasound may also be used. Mammography quantitatively probes the density changes in breast tissue. However, these density changes are not uniquely correlated with the probability of breast cancer. Because of this, mammography serves as a screening technique rather than a diagnostic tool. This is evidenced by the fact that 70-90% of mammographically detected lesions are found to be benign upon biopsy; additionally, mammography fails to detect 20% of all malignant lesions often due to their small size or diffuse nature. The desire to reduce patient trauma, time delay and the high medical costs associated with biopsy has encouraged researchers to develop minimally invasive optical methods for diagnosing malignant lesions in the breast.

A pathologist looks for certain morphological features or patterns to determine malignancy, each of which is associated with biochemical changes in the tissue, figure 1. Many changes in biochemical composition occur between normal tissue and malignant neoplasms. For example, changes in the extracellular matrix during invasion, such as fibrosis, are due to changes in collagen and glycosaminoglycan content. Increased cellular proliferation is accompanied by a decrease in triglycerides and an increase in NADH, flavins, and ATP. Cellular changes, such as differentiation and nuclear pleomorphism, correspond to increased DNA content and concentration. Each of these changes, or possibly a combination of these changes, may provide a spectral marker for identifying pre-malignant lesions and malignant tumors in biological tissue using Raman spectroscopy.

With an appropriate model, Raman spectroscopy can provide quantitative biochemical and morphological information about tissue composition in situ comparable to the information used by a pathologist.1-2 We have previously shown that using a combination of principal component analysis and logistic regression one can distinguish between benign and malignant tumors based on the samples macroscopic Raman spectrum.3 However, the chemical basis for this differentiation remains unknown.

Using our confocal Raman microscope, we are able to collect spectra from individual morphological features. Using this system, we have recently developed a morphological model, shown in figure 2, in an effort to determine the features responsible for the spectral differentiation between benign and malignant breast lesions.4 This model fits macroscopic tissue spectra with a linear combination of basis spectra derived from spectra of the cell cytoplasm, the cell nucleus, fat, β-carotene, collagen, calcium hydroxyapatite, calcium oxalate dihydrate, and cholesterol-like lipid deposits.  Each basis spectrum represents data acquired from multiple patients and, when appropriate, from a variety of normal and diseased states. Modeling is based on the assumptions that the Raman spectrum of a mixture is a linear combination of the spectra of its components and that signal intensity and chemical concentration are linearly related. Least-squares fitting of the macroscopic Raman spectrum of tissue yields the contribution of each basis spectrum to the entire tissue spectrum. To understand the relationship between a tissue sample’s Raman spectrum and its disease state we examine the contribution of each basis spectrum to a variety of pathologies.

Fits of normal, benign, and malignant samples of breast tissue are shown in figure 3.

Using the morphological model, the spectral features of a range of tissue samples can be explained in terms of each sample’s morphological composition. The fit coefficients given by the model normalized to sum to one, represent percentage contributions of chemicals and morphological features to the bulk tissue spectrum. As expected from pathology, when we analyze the fit coefficients of the normal sample, it is primarily composed of fat. The sample diagnosed as fibrosis, a benign condition characterized by scarring, exhibits an increase in the amount of collagen present. Again, the composition is consistent with pathology as scar tissue is formed through collagen accumulation. Furthermore, the fibroadenoma and malignant samples both have a large cell cytoplasm content because they are pathologies which exhibit cellular proliferation.

We are currently in the process of collecting a library of bulk Raman data in order to assess the ability of the morphological model to predict breast tissue disease state.

REFERENCES

1) Manoharan R, Baraga JJ, Feld MS, “Quantitative histochemical analysis of human artery using Raman spectroscopy”, J. Photochem. Photobiol. 16: 211-233 1992.

2) Hanlon EB, Manoharan R, Koo TW, Shafer KE, Motz JT, Fitzmaurice M, Kramer JR, Itzkan I, Dasari RR, Feld MS, “Prospects of in vivo Raman spectroscopy”, Phys. Med. Biol. 45: R1-R59 2000.

3) Manoharan R, Shafer K, Perelman L, Wu J, Chen K, Deinum G, Fitzmaurice M, Myles J, Crowe J, Dasari RR, Feld MS, “Raman spectroscopy and fluorescence photon migration for breast cancer diagnosis and imaging”, Photochem. Photobiol. 67(1): 15-22 1998.

4) Shafer-Peltier K, Haka AS, Fitzmaurice M, Crowe J, Dasari RR and Feld MS “Raman microspectroscopic model of human breast tissue: implications for breast cancer diagnosis in vivo”, J. Raman Spectrosc. in press.

RESEARCH GROUP

▪ Core-Investigator: Michael S. Feld PhD

▪ Graduate Studen: Abigail S. Haka

▪ Collaborators: Joseph Crowe MD, Cleveland Clinic Foundation, Maryann Fitzmaurice

MD, University Hospitals Cleveland/Case Western Reserve University

-----------------------

[pic]

Figure 2. Raman morphological model of breast tissue

[pic]

Figure 1. H&E images of A) normal breast tissue and B) intraductal carcinoma

[pic]

[pic] Figure 3. Examples of morphological fits with corresponding fit contributions. • Data – Model fit (residual plotted below).

................
................

In order to avoid copyright disputes, this page is only a partial summary.

Google Online Preview   Download