Although mammography is the primary clinical imaging modality used to detect breast cancer, limitations in both sensitivity and specificity, particularly in younger and high-risk women, have led to the development of alternative techniques. Overall, mammography has reduced sensitivity in pre-menopausal women [1] and is not clinically advantageous for women under 35 years of age [2]. A general consensus has emerged that mammography is not recommended for women less than 40 years of age, and in the 40 to 50 year old population there is uncertainly regarding its effectiveness. Additional complications arise due to the fact that in pre-menopausal women, mammographic density and false negative rates are greater during the luteal versus follicular phase of the menstrual cycle [3]. Similarly, the use of hormone replacement therapy in post-menopausal women is known to increase mammographic density [4] and has been shown to impede the effectiveness of mammographic screening [5, 6]. In practical terms, up to 10% of all breast cancers, roughly 20,000 cases per year in the US, are not discovered by X-ray mammography [7]. Consequently, new detection technologies are needed that can overcome the limitations of high radiographic density.

The use of near infrared (NIR) optical methods as a supplement to conventional techniques for diagnosing and detecting breast cancer has generated considerable interest. Optical methods are advantageous because they are non-invasive, fast, relatively inexpensive, pose no risk of ionizing radiation, and NIR light can easily penetrate centimeter-thick tissues. Several groups have employed optical methods to measure subtle physiological differences in healthy breast tissue [8–13], to detect tumors [14–22], and to measure tumor response to neoadjuvant chemotherapy [23–25]. Differences in optical signatures between tissues are manifestations of multiple physiological changes associated with factors such as vascularization, cellularity, oxygen consumption, edema, fibrosis, and remodeling.

The primary limitation of optical methods is related to the fact that multiple-scattering dominates NIR light propagation in thick tissues, making quantitative measurements of optical coherence impossible. In this 'diffusion regime', light transport can be modeled as a diffusive process where photons behave as stochastic particles that move in proportion to a gradient, much like the bulk movement of molecules or heat. Quantitative tissue properties can only be obtained by separating light absorption from scattering, typically by using time- or frequency-domain measurements and model-based computations [26–29]. The underlying physical principle of these 'photon migration' methods is based on the fact that the probability of light absorption (i.e. molecular interactions) is 50 to 100-fold lower than light scattering due to dramatic differences in tissue scattering versus absorption lengths [30, 31].

Quantitative diffuse optical methods can be used in breast diagnostics to form images (diffuse optical imaging (DOI)) and obtain spectra (diffuse optical spectroscopy (DOS)). DOI and DOS are conceptually similar to the relationship between magnetic resonance imaging (MRI) and magnetic resonance spectroscopy. In general, DOI is used to form images of subsurface structures by combining data from a large number of source-detector 'views' (i.e. in planar or circular transmission geometry) using inverse tomographic reconstruction techniques [32]. DOI typically utilizes a limited number of optical wavelengths (e.g., two to six) and a narrow temporal bandwidth. In contrast, DOS employs a limited number of source-detector positions (e.g., one to two) but employs broadband content in temporal and spectral domains (i.e., hundreds of wavelengths) to recover complete absorption and scattering spectra from approximately 650 to 1,000 nm. Although an ideal DOI design would employ hundreds or thousands of source-detector pairs and wavelengths, several engineering considerations related to measurement time currently limit the practicality of this approach.

A substantial body of work has emerged over the past decade that demonstrates how tomographically based DOI methods can accurately localize subsurface structures. Optimal clinical decision-making, however, requires understanding the precise biochemical composition or 'fingerprint' of these localized inhomogeneities. This information can be obtained by fully characterizing the spectral content of breast tumors using quantitative DOS. DOS signatures are used to measure tissue hemoglobin concentration (total, oxy-, and deoxy-forms), tissue hemoglobin oxygen saturation (oxy-hemoglobin relative to the total hemoglobin), water content, lipid content and tissue scattering. Several research groups have demonstrated the sensitivity of these tissue components to breast physiology and disease [8, 10, 11, 33]. Critical challenges remain to determine the precise relationship between these quantitative measures and cancer. Consequently, this paper reviews our efforts to determine tumor biochemical composition from low-resolution spatial maps of broadband absorption and scattering spectra.

To minimize partial volume sampling effects and attribute our signals specifically to breast tumors despite high mammographic density, we have studied 12 pre-menopausal 30 to 39 year old subjects with locally advanced, stage III invasive disease, focusing on the question, "what do tumors 'look' like?" Because the biological processes that determine the origins of optical contrast are conserved across spatial scales, intrinsic optical signals measured from these subjects are expected to be similar for earlier stage disease. We highlight this population because conventional methods are generally considered to be ineffective in younger women. We also present results of DOS measurements during neoadjuvant chemotherapy to demonstrate the sensitivity of optics to physiological perturbations within one week of treatment. Thus, these studies provide critical information regarding the spectral content of DOI necessary for clinical applications, such as early cancer detection, distinguishing between malignant and benign tumors, and monitoring the effects of neoadjuvant chemotherapy.

Broadband DOS measurements were made with the laser breast scanner (Fig. 1a). The laser breast scanner is a bedside-capable system that combines frequency-domain photon migration with steady-state tissue spectroscopy to measure complete (broadband) NIR absorption and reduced scattering spectra of breast tissue in vivo. Detailed descriptions of the instrumentation and theory have been provided elsewhere [34–36].

DOS measurements are made by placing the hand-held probe (Fig. 1b) on the tissue surface and moving the probe to discrete locations along a line at 1.0 cm intervals. This forms a linescan across the lesion and surrounding normal tissue (Fig. 2a). The number of DOS positions varies depending on the lesion size. For comparison, a linescan is also performed at an identical location on the contralateral breast. Two measurements are made in each location and all measurement positions are marked on the skin with a surgical pen. The average laser optical power launched into the tissue is about 10 to 20 mW and the total measurement time to generate complete NIR absorption and scattering spectra from a single position is typically about 30 seconds. A complete DOS study including calibration time is approximately 30 to 45 minutes.

The probe source and detector separation is 28 mm, from which we estimate a mean penetration depth of approximately 10 mm in the tissue. The actual tissue volume interrogated, which is determined by multiple light scattering and absorption (Fig. 2b), extends above and below the mean penetration depth and is estimated to approximately 10 cm³.

Laser breast scanner measurements produce complete absorption and reduced scattering spectra across the NIR (650 to 1,000 nm) at each probe position. From the absorption spectrum, quantitative tissue concentration measurements of oxygenated hemoglobin (ctO₂Hb), deoxygenated hemoglobin (ctHHb), water (ctH₂O), and lipid are calculated [8]. From these parameters total tissue hemoglobin concentration (ctTHb = ctO₂Hb + ctHHb) and tissue hemoglobin oxygenation saturation (stO₂ = ctO₂Hb/ctTHb × 100%) are calculated. A tissue optical index (TOI) was developed as a contrast function by combining DOS measurements; TOI = ctHHb × ctH₂O/(%lipid). The parameters of this contrast function were determined from an evaluation of DOS measurements in a larger population of 58 malignant breast lesions [37]. Spatial variations in TOI allow us to rapidly locate the maximum lesion optical contrast. Tissue scattering is reported by the results of a power law fit of the form scattering = Aλ^-SP, where λ is the optical wavelength and SP is scatter power [38, 39]. Data were analyzed with custom software developed in Matlab (MathWorks, Inc., Natick, MA, USA).

Tumor and normal breast tissues displayed significant differences in ctHHb (p = 0.005), ctO₂Hb (p = 0.002), ctH₂O (p = 0.014), and lipids (p = 0.0003) in a population of 12 women aged from 30 to 39 years. These physiological data were assembled into a TOI to enhance the functional contrast between malignant and normal tissues; however, stO₂ was not found to be a reliable index in this regard. A 50% decrease in TOI was measured within 1 week for a patient undergoing neoadjuvant chemotherapy.

DOS and DOI are relatively inexpensive technologies that do not require compression, are intrinsically sensitive to the principal components of breast tissue, and are compatible with the use of exogenous molecular probes. DOS is easily integrated into conventional imaging approaches such as MRI, ultrasound, and mammography; and performance is not compromised by structural changes that impact breast density. As a result, diffuse optics may be advantageous for populations with dense breasts, such as younger women, high-risk subjects, and women receiving hormone replacement therapy. Because NIR light is non-ionizing, DOI can be used to monitor physiological changes on a frequent basis without exposing the tissue to potentially harmful radiation. Finally, because DOS can be used to quantitatively assess tumor biochemical composition, it can be applied to monitoring tumor response to therapy. Because these changes occur predominantly early in the course of treatment, we anticipate that diffuse optics will play an important role in minimizing toxicity, predicting responders early in the course of therapy, and developing 'real time' strategies for individualized patient care.

This article is part of a review series on Imaging in breast cancer, edited by David A Mankoff.

Other articles in the series can be found online at http://breast-cancer-research.com/articles/review-series.asp?series=BCR_Imaging