Phase-Contrast Diffuse Optical Tomography

Phase-Contrast Diffuse Optical Tomography: Pilot Results in the Breast1 Xiaoping Liang, MS, Qizhi Zhang, PhD, Changqing Li, PhD, Stephen R. Grobmyer, MD Laurie L. Fajardo, MD, Huabei Jiang, PhD

Rationale and Objectives. We sought to investigate the utility of phase-contrast diffuse optical tomography (PCDOT) for differentiation of malignant and benign breast masses in humans and to compare PCDOT with conventional diffuse optical tomography (DOT) for analysis of breast masses in humans. Materials and Methods. Thirty-five breast masses were imaged in 33 patients (mean age, 51 years; range, 22– 80) using PCDOT. Images characterizing the tissue refractive index, and absorption and scattering coefficients of breast masses were obtained with a finite element-based reconstruction algorithm. Theses images were then analyzed and compared with the biopsy/pathology results for all the cases examined. Results. Malignant lesions tended to have a decreased refractive index, allowing them to be discriminated from benign lesions in most cases, whereas absorption and scattering images were unable to accurately discriminate benign from malignant lesions. The sensitivity, specificity, false-positive value, and overall accuracy for refractive index imaging were 81.8%, 70.8%, 29.2%, and 74.3%, respectively. The accuracy of refractive index imaging increases with increasing patient age. Conclusion. Refractive index is a new parameter for optical imaging that may be helpful in differentiating between malignant and benign masses in the breast. Key Words. Absorption coefficient; scattering coefficient; refractive index; phase-contrast; diffuse optical tomography. ©

AUR, 2008

The low specificity of x-ray mammography, the current clinical tool for breast imaging, suggests that many women without cancer will receive recommendations to undergo a breast biopsy (ie, false-positive examinations). This results in four or five benign breast biopsy results for every one malignancy that is detected (1,2). In addi-

Acad Radiol 2008; 15:859 – 866 1

From the J. Crayton Pruitt Family Department of Biomedical Engineering, University of Florida, 130 BME Building/PO Box 116131, Gainesville, FL 32611 (X.L., Q.Z., C.L., H.J.); Department of Surgery, University of Florida, Gainesville, FL 32610 (S.R.G.); and Department of Radiology, University of Iowa, Iowa City, IA, 52242 (L.L.F.). This work was supported in part by a grant from the National Institutes of Health (R01 CA090533). Received August 28, 2007; accepted January 22, 2008. Address correspondence to: H.J. e-mail: [email protected]

© AUR, 2008 doi:10.1016/j.acra.2008.01.028

tion, mammography has an unacceptable false-negative rate for patients with radiodense breast tissues (3– 6). For example, the sensitivity of screening mammography decreases from a high of 88% in the predominantly fatty breast to 62% in the dense breast (3). Thus, there is a critical need to develop new technologies that can improve the sensitivity and specificity of mammography. Several other conventional techniques are currently under investigation for breast cancer detection, including ultrasound (US), magnetic resonance imaging (MRI), and computed tomography (CT) (7). Among these conventional alternatives or supplements, MRI is the most widely studied and potentially promising solution to dense breast imaging. A recent MRI study of 821 women showed a sensitivity of 88% and corresponding specificity of 68% regardless of breast density (8). Similar high sen-

859

LIANG ET AL

Academic Radiology, Vol 15, No 7, July 2008

Table 1 Characteristics of Malignant Lesions: Pathology, Mammography, and Phase-contrast Diffuse Optical Tomography The Difference of RI Between Lesion and Normal Tissue

BIRADS Score* Biopsy Findings

Lesion Size† (cm)

Lobular carcinoma (n ⫽ 3) Infiltrating ductal carcinoma (n ⫽ 6) Extensive intraductal breast carcinoma (n ⫽ 1) Invasive carcinoma with mixed ductal and lobular features (n ⫽ 1)

0.5, 1.0, 2.0 5, 2, 2, 10, 1.9 2.5

2

3

4

5

⌬n ⬍ 0

⌬n ⬎ 0

1 3

1 1

3 4 1

0 2 0

1

0

1

0.8

1

BIRADS, Breast Imaging Reporting and Data System; RI, refractive index. *BIRADS score was not available to some mammograms. †Lesion size was not available to some patients.

sitivities of MRI for breast cancer have also been found in other studies involving targeted imaging of high-risk populations (9). Although MRI appears to offer high sensitivity in the dense breast, its specificity has been far more modest. In addition, MRI is costly (10,11) and not suitable for breast screening. Near-infrared (NIR) light– based diffuse optical tomography (DOT) is emerging as a nonconventional alternative for breast cancer detection. The ability of DOT techniques to noninvasively image and analyze both tissue structure and function, as well as the advantages of low cost and portability, has great potential to make DOT an ideal candidate for routine breast screening. Although DOT has been shown to have high sensitivity for breast tumor detection, its specificity is limited (12–22). Detection of tumors by DOT is so far primarily based on tissue absorption and scattering parameters or absorption-derived functional parameters. However, these imaging parameters available from the current DOT do not appear to be able to fully characterize breast tissue, which results in limited sensitivity and specificity. We moved a step forward in 2003 to show for the first time that refractive index/phase contrast could be used as a new imaging parameter for NIR tomography (23). It is known that the refractive index (RI) in tissue depends on the tissue’s physical and chemical properties. It is also known that the RI and scattering coefficient are correlated in a complicated manner and that scattering coefficient is a consequence of the local RI distribution (24,25). Interestingly, although RI as an image contrast remains unexplored in DOT, it has been widely used for many decades in optical microscopy (26,27). There is also a growing interest in developing phase-contrast CT for better imag-

860

ing soft tissues or weakly absorbing materials (28,29). Local variation in RI gives rise to contrast in images obtained with optical coherent tomography (OCT) (30). In the NIR region, the recent OCT measurements of RI in animal breast cancer models (31) and the previous study of two normal/tumor human breast tissue samples (32) indicate that cancer and normal tissues demonstrate significant differences in their refractive indexes. Here, we report pilot clinical data on the use of phasecontrast DOT (PCDOT) for breast cancer detection. Images characterizing the tissue RI and the absorption and scattering coefficients were obtained with a finite elementbased reconstruction algorithm. Our results are compared to biopsy, mammography, and US information. The accuracy of RI images was analyzed relative to tumor size, patient age, and breast size. The results suggest that the RI may be a useful parameter to differentiate malignant from benign breast tumors.

MATERIALS AND METHODS Patient Examinations The study was approved by the institutional review board, and the study was conducted in full compliance with the accepted standards for research involving human subjects. Signed informed consent was obtained from all study participants. Thirty-five breasts of 33 different patients were imaged as a part of this study. Biopsy results (indicated in Tables 1 and 2) demonstrated 11 invasive carcinomas (malignant group) and 24 benign lesions (benign group). Malignant lesions included lobular breast carcinoma, ductal carcinoma, and extensive intraductal

PCDOT: RESULTS IN THE BREAST

Academic Radiology, Vol 15, No 7, July 2008

Table 2 Characteristics of Benign Lesions as Determined at Biopsy, Mammography, and Phase-contrast Diffuse Optical Tomography The difference of RI Between Lesion and Normal Tissue

BIRADS Score* Imaging Findings Mass

Asymmetric density

Calcifications

Pathology Findings

Lesion Size† (cm)

Adipose tissue with focal fibrosis (n ⫽ 1) Mass or lump (n ⫽ 4) Cysts (n ⫽ 6) Lymph node (n ⫽ 1) Fibroadenomas (n ⫽ 2) Infarcted intraductal papilloma (n ⫽ 1) Fibrocystic mastophy (n ⫽ 1) Fibroadipose tissue (n ⫽ 5) Fibroadenomatoid change and microcalcifications (n ⫽ 1) Microcalcification, atypical ductal hyperplasia present (n ⫽ 1) Benign calcification (n ⫽ 1)

N/A 1.3, 2 1.6, 0.3, 0.8 2.1 2, 3 1 2.3 1.8, 0.8 N/A N/A 0.3

1

2

2 1 1

2

3

4

2 1 1 1 1 3 1

1

⌬n ⬍ 0

⌬n ⬎ 0

0 1 1 0 2 0 0 1 1 1 0

1 3 5 1 0 1 1 4 0 0 1

BIRADS, Breast Imaging Reporting and Data System; RI, refractive index. *BIRADS score was not available to some mammograms. †Lesion size was not available to some patients.

breast carcinoma; benign lesions were cysts, fibroadenoma, fibroadipose tissue, fibroadenomatoid change and microcalcifications, infarcted intraductal papilloma, and lymph nodes. Twenty-eight patients were examined using a compact, parallel-detection diffuse optical mammographic system (Fig. 1), whereas the remainder (n ⫽ 5) were imaged with a similar system described in detail elsewhere (13). One feature of the system is that all the mechanical, electronic, and optical components are arranged in a single enclosed frame under the examination table (Fig. 1). Figure 2 shows the breast–fiberoptic interface, a coneshaped frustum containing four layers of fiberoptic bundles whose diameter is adjustable. In each layer there are 16 detection and 16 source fibers positioned alternatively with equal spacing in a circular array (the coronal two-dimensional images presented here were obtained from one of the four layers; see, for example, the dashed circle in Fig. 2b). Positions are adjustable from 4.0 to 15 cm in circumference. During the optical examination, the patient was positioned prone on the table and the breast to be imaged was placed pendant through an opening where the fiberoptic array was in gentle contact with the breast (see Fig. 2a). The examination table can be translated vertically, which allows for imaging of the breast from the chest wall to near the nipple. Laser light at 785 nm is sequentially delivered to the breast. Light multiplexing is accomplished by two high-precision moving stages. A rotary stage couples the laser light to the

Figure 1. Photograph of phase contrast diffuse optical mammographic system.

16 source fibers, and a translation stage relays the light between the four layers. The translation stage also couples the received light from one layer to another. Sixteen photomultiplier tubes with computer-controlled gains record the detected signal. The entire system is controlled by a LabVIEW program (National Instruments, Austin, TX). A detailed description of the system was given previously (33). Image Reconstruction Methods The spatial distribution of the tissue RI is reconstructed using a regularized nonlinear iterative algorithm based on the iterative finite element solution to the following diffusion equation considering spatially varying RI (34).

861

LIANG ET AL

Academic Radiology, Vol 15, No 7, July 2008

Figure 2. Photograph (left) and schematic (right) of the fiberoptic array/probe.

ⵜ · D ⵜ ⌽(r) ⫹

2D n

ⵜ n · ⵜ ⌽(r) ⫺ ␮a⌽(r) ⫽ ⫺S0␦(r ⫺ r0) (1)

where ⌽ is the photon density; n is the refractive index; D and ␮a are the diffusion and absorption coefficients, respectively; and S0 is the source strength. For PCDOT, the reconstruction algorithm is similar to that of conventional DOT (18,35); that is, we use a regularized Newton’s method to update an initial RI distribution iteratively to minimize an object function composed of a weighted sum of the squared difference between computed and measured optical data at the surface of the medium. The computed optical data (i.e., photon density) are obtained by solving Equation 1 with the finite element method. Tissue absorption and diffusion or scattering coefficients were reconstructed separately with conventional DOT. The algorithm has been successfully tested using extensive tissue-mimicking phantom measurements with different contrasts of RI between the target and background (36). Statistical Analysis For statistical analysis, sensitivity was calculated as TP/(TP ⫹ FN), specificity as TN/(TN ⫹ FP), FPR (rate of false positive) as (1 – specificity), and overall accuracy as (TP ⫹ TN)/(TP ⫹ TN ⫹ FP ⫹ FN), where TP, TN, FP, and FN are, respectively, the number of true-positive,

862

true-negative, false-positive, and false-negative findings. Information obtained at biopsy and at 2 years of follow-up mammography (for benign cases) was used as the reference standard.

RESULTS Reconstructed RI and absorption and scattering images (coronal plane) for two representative cases are shown in Figures 3 and 4. The first case is a 52-year-old woman with a 3-cm infiltrating ductal carcinoma. The patient’s mammograms are shown in Figures 3a and 4a. The RI image (Fig. 3b) exhibits marked decrease in RI in the region of the tumor (indicated by arrow, about 6 o’clock), whereas both the absorption and scattering images (Figs. 3c and 3d) show a clear increase in absorption and scattering coefficient in the tumor area. The second case is a 64-year-old woman with benign microcalcifications. We note that an increased RI (Fig. 4b) and increased absorption and scattering coefficients (Figs. 4c and 4d) are present in the area of the microcalcifications (indicated by arrow). Also, tissue RI is not significantly dependent on wavelength and the RI images we reconstructed at several different wavelengths give almost identical results. The characteristic feature of decreased RI for cancer and increased RI for benign lesions appears to be true for most cases studied. To confirm this observation, we calculated the difference in RI values between the lesion and

Academic Radiology, Vol 15, No 7, July 2008

PCDOT: RESULTS IN THE BREAST

Figure 3. A 52-year-old woman with infiltrating ductal carcinoma. (a) Mediolateral mammogram. (b) Refractive index image. (c, d) Absorption and scattering coefficient images. The axes (left and bottom) illustrate the spatial scale (millimeters), whereas the color scale (right) records the normalized refractive index (dimensionless), absorption, or scattering coefficient (mm⫺1).

the surrounding normal tissue for all the cases and present it in Tables 1 and 2 for malignant and benign lesions, respectively. For 9 of 11 malignant lesions (Table 1), ⌬n ⬍ 0, where ⌬n is the difference in RI value between the tumor and its surrounding, whereas for 17 of 24 benign cases (Table 2), ⌬n ⬎ 0. We also note that the absorption and scattering images do not have such characteristic features that allow us to distinguish between malignant and benign lesions. However, benign cystic lesions do appear to have different absorption and/or scattering characteristics compared to solid lesions that allow them to be differentiated using absorption and scattering images (Fig. 5a), a finding consistent with our previous studies (37). The characteristic features seen in the RI images are further confirmed in Figure 5b and 5c, which graphically shows that the malignant and benign groups can be separated, especially when the RI information is combined with the absorption and/or scattering parameters. Based on the results shown in Tables 1 and 2 and

Figure 5, the sensitivity, specificity, false-positive value, and overall accuracy were calculated and are given in Table 3. The performance of PCDOT over patient age, lesion size, and breast size (based on the radius of image plane) is shown in Table 4. We note that the accuracy is not significantly different for lesions larger than or smaller than 1 cm. We do note that the accuracy is greater with increased patient age. For women younger than 40 years, the accuracy is 0.5714, whereas for women older than 60 years, the accuracy reaches 0.9. The effect of breast size on accuracy is variable. The highest accuracy, 0.83, is observed when the breast size was between 40.5 and 59 mm. These figures are not statistically significant, given the limited number of lesions in each category. The relationship of the accuracy versus the breast size is not linear, because the smaller breasts may have smaller lesions or are highly heterogeneous, resulting in lower signal-tonoise ratio (and thus lower accuracy).

863

LIANG ET AL

Academic Radiology, Vol 15, No 7, July 2008

Figure 4. A 64-year-old woman with benign microcalcifications. (a) Mediolateral mammogram. (b) Refractive index images. (c, d) Absorption and scattering coefficient images. The axes (left and bottom) illustrate the spatial scale (millimeters), whereas the color scale (right) records the normalized refractive index (dimensionless), absorption, and scattering coefficient, respectively (mm⫺1).

Figure 5. (a) Absorption (⌬a) and scattering (⌬s) value relationship. (b) Absorption (⌬a) and refractive index (⌬n) value relationship. (c) Scattering (⌬s) and refractive index (⌬n) value relationship. Here, if the value of absorption, scattering, or refractive index (⌬n) in the lesion is larger than that in the surrounding tissue, ⌬a, ⌬s or ⌬n ⫽ 1; otherwise ⌬a, ⌬s or ⌬n ⫽ ⫺1.

DISCUSSION In this study, we found that the malignant lesions generally show a decreased RI, whereas the benign lesions consistently exhibit an increased RI relative to the surrounding normal tissue. We also found that the absorption and scattering images from the conventional DOT do not show such a characteristic feature. Although a systematic

864

study of the physiologic change causing the contrast of RI is a subject of future investigation, initial results using animal models suggest that the glucose consumption in malignant tumors is significantly elevated compared to the surrounding normal tissue (38), possibly resulting in decreased RI in the tumors (lower glucose concentration gives lower RI) (39). In some cases, if necrosis presents in the tumors, a decreased RI can also be expected, ac-

PCDOT: RESULTS IN THE BREAST

Academic Radiology, Vol 15, No 7, July 2008

Table 3 Differentiation of Benign from Malignant Tumors Based on Refractive Index Images True Positives

True Negatives

False Positives

False Negatives

Sensitivity

Specificity

FPR

Overall Accuracy

9

17

7

2

81.8%

70.8%

29.2%

74.3%

Table 4 Accuracy of Phase-contrast Diffuse Optical Tomography According to Tumor Size, Age, or Size of Breast (Imaging Plane)

Lesion size (cm) ⬎1 ⬍1 Unknown Age (yr) ⬍40 41–49 50–59 ⬎60 Size of breast (radius of image plane) (mm) ⬍40 40.5–59 ⬎59.5

No. of Cases

True

False

20 10 5

15 8 3

5 2 2

7 8 10 10

4 5 8 9

3 3 2 1

0.5714 (4/7) 0.625 (5/8) 0.8 (8/10) 0.9 (9/10)

17 12 6

12 10 4

5 2 2

0.7059 (12/17) 0.8333 (10/12) 0.6667 (4/6)

Accuracy

0.75 (15/20) 0.8 (8/10) 0.6 (3/5)

cording to the recent study of breast cancer animal models (31). The recent in vivo work of Zysk et al. (31) indicates that the RI of tumor tissue is 1.39, which is 5% smaller than its surrounding normal tissue (RI ⫽ 1.47). This level of contrast is high enough to be sensitively detected by DOT, because we have previously shown, using tissue-like phantom experiments, that even 1% contrast between the target and background could be quantitatively detected using our PCDOT approach (39).

CONCLUSIONS We presented a pilot study aiming to initially evaluate the PCDOT approach in the breast with malignant and benign lesions. This study of 35 breast masses shows that malignant and benign breast lesions demonstrate characteristic differences in RI value. Although we have demonstrated the use of PCDOT for breast cancer detection, the approach described can be easily adapted for imaging of other cancers. For example, if small fiberoptic probes are

used, PCDOT can be integrated with an endoscope for gastrointestinal cancer diagnosis/prognosis, including colon and esophagus cancers. Small fiberoptic probe-based PCDOT can certainly be used for skin, oral, and neckrelated cancer diagnosis as well. REFERENCES 1. Doyle AJ, Murray KA, Nelson EW. Selective use of image-guided largecore needle biopsy of the breast: Accuracy and cost-effectiveness. AJR Am J Roentgenol 1995; 165:281–284. 2. Liberman L, Feng TL, Dershaw DD. US-guided core breast biopsy: Use and cost-effectiveness. Radiology 1998; 208:717–723. 3. Carney PA, Miglioretti DL, Yankaskas BC, et al. Individual and combined effects of age, breast density, and hormone replacement therapy use on the accuracy of screening mammography. Ann Intern Med 2003; 138:168 –175. 4. Sickles EA, Miglioretti DL, Ballard-Barbash R, et al. Performance benchmarks for diagnostic mammography. Radiology 2005; 235:775– 790. 5. Boyd NF, Lockwood GA, Byng JW, et al. Mammographic densities and breast cancer risk. Cancer Epidemiol Biomarkers Prev 1998; 7:1133– 1144. 6. Lam PB, Vacek PM, Geller BM, et al. The association of increased weight, body mass index, and tissue density with the risk of breast carcinoma in Vermont. Cancer 2000; 89:369 –375. 7. Sabel M, Aichinger H. Recent developments in breast imaging. Phys Med Biol 1996; 41:315–368. 8. Bluemke DA, Gatsonis CA, Chen MH, et al. Magnetic resonance imaging of the breast prior to biopsy. JAMA 2004; 292:2735–2742. 9. Warner E, Plewes DB, Hill KA, et al. Surveillance of BRCA1 and BRCA2 mutation carriers with magnetic resonance imaging, ultrasound, mammography, and clinical breast examination. JAMA 2004; 292:1317–1325. 10. Bluemke DA, Gatsonis CA, Chen MH, et al. Magnetic resonance imaging of the breast prior to biopsy. JAMA 2004; 292:2735–2742. 11. Berg WA, Gutierrez L, NessAiver MS, Carter WB, Bhargavan M, Lewis RS, Loffe OB. Diagnostic accuracy of mammography, clinical examination, US, and MR imaging in preoperative assessment of breast cancer. Radiology 2004; 233:830 – 849. 12. Ntziachristos V, Yodh AG, Schnall M, Chance B. Concurrent MRI and diffuse optical tomography of breast after indocyanine green enhancement. Proc Natl Acad Sci USA 2000; 97:2767–2772. 13. Jiang H, Iftimia N, Xu Y, Eggert J, Fajardo LL, Klove K. Near-infrared optical imaging of the breast with model-based reconstruction. Acad Radiol 2002; 9:186 –194. 14. Zhu Q, Cronin EB, Currier AA, et al. Benign versus malignant breast masses: Optical differentiation with US-guided optical imaging reconstruction. Radiology 2005; 237:57– 66. 15. Cerussi AE, Berger AJ, Bevilacqua F, et al. Sources of absorption and scattering contrast for near-infrared optical mammography. Acad Radiol 2001; 8:211–218. 16. Durduran T, Choe R, Culver JP, et al. Bulk optical properties of healthy female breast tissue. Phys Med Biol 2002; 47:2847–2861. 17. Ntziachristos V, Yodh AG, Schnall MD, et al. MRI-guided diffuse optical spectroscopy of malignant and benign breast lesions. Neoplasia 2002; 4:347–354.

865

LIANG ET AL

18. Jiang H, Xu Y, Iftimia N, Eggert J, Klove K, Baron L, Fajardo L. Threedimensional optical tomographic imaging of breast in a human subject. IEEE Trans Med Imaging 2001; 20:1334 –1340. 19. Brooksby B, Jiang S, Dehghani H, Pogue BW, Paulsen KD. Combining near-infrared tomography and magnetic resonance imaging to study in vivo breast tissues: Implementation of a Laplacian-type regularization to incorporate magnetic resonance structure. J Biomed Optics 2005; 1–10. 20. Spinelli L, Torricelli A, Pifferi A, et al. Bulk optical properties and tissue components in the female breast from multiwavelength time-resolved optical mammography. J Biomed Optics 2004; 9:1137–1142. 21. Franceschini MA, Moesta KT, Fantini S, et al. Frequency-domain techniques enhance optical mammography: Initial clinical results Prod Natl Acad Sci USA 1997; 94:6468 – 6473. 22. Grosenick D, Moesta KT, Wabnitz H, et al. Time-domain optical mammography: Initial clinical results on detection and characterization of breast tumors. Appl Optics 2003; 42:3170 –3186. 23. Jiang H, Xu Y. Phase-contrast imaging of tissue using near-infrared diffusing light. Med Phys 2003; 30:1048 –1051. 24. Knuttel JM, Boehlau-Godau M. Spatially confined and temporally resolved refractive index and scattering evaluation in human skin performed with optical coherence tomography. J Biomed Opt 2000; 5:83–92. 25. Schmitt JM, Kumar G. Optical scattering properties of soft tissue: A discrete particle model. Appl Opt 1998; 37:2788 –2797. 26. Bennett H, Osterberg H, Jupnik H, Richards O. Phase Microscope: Principles and Applications. New York: John Wiley & Sons, 1951. 27. Bereiter-Hahn J, Fox C, Thorell B. Quantitative reflection contrast microscope of living cells. J Cell Biol 1979; 82:767–779. 28. Takeda T, Momose A, Hirano K, Haraoka S, Watanabe T, Itai Y. Human carcinoma: Early experience with phase-contrast X-ray CT with

866

Academic Radiology, Vol 15, No 7, July 2008

29.

30.

31. 32.

33. 34.

35. 36. 37.

38.

39.

synchrotron radiation: Comparative specimen study with optical microscope. Radiology 2000; 214:298 –301. Davis J, Gao D, Gureyev T, Stevenson A, Wilkins S. Phase-contrast imaging of weakly absorbing materials using hard X-rays. Nature 1995; 373:595–598. Tearney GJ, Brezinski ME, Southern J, Bouma B, Hee M, Fujimoto J. Determination of the refractive index of highly scattering human tissue by optical coherence tomography. Opt Lett 1995; 20:2258 –2260. Zysk A, Chaney E, Boppart S. Refractive index of carcinogen-induced rat mammary tumors. Phys Med Biol 2006; 51:2165–2177. Das B, Liu F, Alfano R. Time-resolved fluorescence and photon migration studies in biomedical and model random media. Rep Prog Phys 1997; 60:227–292. Iftimia N, Gu X, Xu Y, Jiang H. A compact, parallel-detection diffuse optical mammography system. Rev Sci Instr 2003; 74:2836 –2842. Jiang H. The diffusion approximation for turbid media with a spatially varying refractive index. Proc Adv Opt Imaging Photon Migration Opt Soc Am 2000; 366 –368. Iftimia N, Jiang H. Quantitative optical image reconstruction of turbid media using DC measurements. Appl Opt 2000; 39:5256 –5261. Jiang H, Xu Y. Phase-contrast imaging of tissue using near-infrared diffusing light. Med Phys 2003; 30:1048 –1051. Gu X, Zhang Q, Bartlett M, Schutz L, Fajardo LL, Jiang H. Differentiation of cysts from solid tumors in the breast with diffuse optical tomography. Acad Radiol 2004; 11:53– 60. Casciari JJ, Sotirchaos SV, Sutherland RM. Variations in tumor cell growth rates and metabolism with oxygen concentration, glucose concentration, and extracellular pH. J Cell Physiol 2005; 151:386 –394. Liang X, Zhang Q, Jiang H. Quantitative reconstruction of refractive index distribution and imaging of glucose concentration by using diffusing light. Appl Opt 2006; 45:8360 – 8365.