Breast conserving surgery (BCS) is a recommended treatment for early-stage breast cancer and for breast cancers that have been reduced in size by neoadjuvant therapy. The goal of BCS is to excise the tumor along with a margin of normal tissue, while preserving as much of the normal breast tissue as possible. Unfortunately, as many as 18–72% of patients undergoing BCS require repeat surgeries due to a close or positive surgical margin diagnosed post-operatively and thus, require a re-excision surgery to achieve cancer free margins 
. These re-excision surgeries are not only a burden to patients financially but also physically and psychologically and can delay recommended adjuvant therapies. Additionally, 10–36% of women requiring re-excision will undergo mastectomy which significantly alters a patient’s initial treatment decision 
. The large variation in re-excisions is thought to be due to differences in surgeon’s training, in the definition of a close margin, and in the perceived risk of focally positive margins versus extensive involvement 
Histopathology is the current gold standard for determining surgical margin status. At many hospitals, including Duke University Medical Center (DUMC), the standard of care is to grossly section the specimen into 3 mm slices perpendicular to the long axis of the specimen. The tissue slices are then further sectioned and from each of the resulting paraffin blocks, a 5 µm thick section is taken for staining and histological review. The pathologic margin status is an important predictor of local recurrence of an invasive or in situ
cancer after BCS 
. Thus, re-excision of the tumor margin is essential to reduce the risk of local recurrence 
. In post-operative pathology, it is not feasible to section and analyze the entire specimen, especially when the specimens are large. This issue was evaluated by Guidi et al. 
where they looked at the presence of tumor in perpendicularly sliced sections of the inked margin (the type of analysis done here at DUMC) versus evaluating tissue from an en face
cut of the margin (i.e. a shaved margin). They found that of 69 positive shaved margins, only 42 inked margins were found to be positive, indicating that residual carcinomas may be missed with the current approach of sampling tissue every couple of millimeters.
A small number (less than 5%) of hospitals that perform BCS currently utilize intra-operative cytologic or pathologic analysis of tumor margins. Touch-preparation (touch-prep) cytology is a technique in which cells on the surface of the tissue are transferred to glass slides by touching the specimen to the glass, and are then stained for pathologic observation. For frozen section analysis, the tissue is frozen and select microscopically thin sections are cut from the specimen for pathologic observation. Typically a much smaller fraction of the tumor margin is sampled in frozen section than in post-operative pathology. Touch-prep cytology and frozen-section analysis can reduce surgical re-excision rates; reported sensitivities and specificities for touch-prep are 38–100% and 83–100%, respectively 
. Sensitivity of frozen section ranges from 59–91% and specificity ranges from 86–100% 
. Although these two approaches have been shown to be beneficial to the surgeon, there are a number of limitations with each. Both procedures are time consuming and require special expertise by a pathologist at the time of surgery. Additionally, touch-prep cytology allows for the evaluation of the whole lumpectomy surface but is not capable of detecting close margins since only cells at the specimen surface are sampled. Frozen section analysis may not be utilized on every patient but may be determined in collaboration with the surgeon, pathologist, and radiologist after a laborious process of gross examination and specimen mammography 
. Sampling issues are also a problem since the entire specimen cannot be evaluated.
The above discussion points to the fact that surgery to remove the cancer and obtain clear margins is a collaborative effort between the surgeon and the pathologist (and in some institutions, the radiologist). In spite of this, there can be substantial variability in the prediction of positive margins in the intra-operative and post-operative settings. Surgeons do not have adequate intra-operative assessment tools to ensure that the cancer has been completely removed at the time of first surgery. Pathologists do not have adequate tools for sampling from areas on large tumor margins. The lack of these capabilities represents a significant unmet clinical need for margin assessment for both the surgeon and pathologist.
Optical imaging of tissue is an attractive solution to this problem because it is relatively fast and non-destructive. Optical techniques can also measure features related to the histological landscape without the need for labels. provides a breakdown of the different optical tools that have been leveraged to measure breast tissue constituents for different applications in breast cancer ranging from diagnostic biopsy to margin assessment to monitoring of response to neoadjuvant therapy. This table shows that no matter what tool is used, the primary sources of contrast in breast tissue are scattering (which primarily reflects the fibroglandular content of breast tissue), lipid and carotenoid concentration (which reflects the fatty content of the breast tissue content), hemoglobin (which reflects tissue vascularity), and in the case of fluorescence, metabolism of the tumor cells.
Optical sources of contrast in breast tissue.
Pioneering optical studies to characterize breast tumor margins was carried out by Bigio et al 
where they used reflectance spectroscopy in the UV-Visible range to look at sites within the tumor bed in 24 patients (13 cancer and 59 normal sites). This work was important in that it represented initial evidence of absorption and/or scattering contrast in residual breast cancer. Keller et al published on diffuse reflectance and fluorescence spectroscopy to detect cancerous sites on excised breast tumor margins in 32 patients (145 normal and 34 individual tumor sites), and reported a sensitivity and specificity of 85% and 96%, respectively, for classifying individual sites (not margins) 
. Haka et al published on Raman spectroscopy of tumor sites on freshly sliced lumpectomy specimens in 21 patients (123 benign and 6 malignant tissue sites) and exploited fat and collagen contrast to achieve sensitivity and specificity of 83% and 93%, respectively for classifying individual sites 
. Nguyen et al 
demonstrated that optical coherence tomography detects ex vivo
margin positivity in 20 patients (11 positive/close margins and 9 negative margins), with sensitivity and specificity of 100% and 82%, respectively by exploiting scattering associated with increased cell density. Nachabe et al 
used diffuse reflectance spectroscopy to acquire spectra from 102 ex vivo
samples that consisted of adipose, glandular, fibroadenoma, invasive carcinoma, and DCIS. Using a K-nearest neighbor algorithm, malignant and non-malignant samples were separated with a sensitivity of 94±4% and a specificity of 98±2%.
We published recently on using a quantitative diffuse reflectance spectral imaging technique to non-destructively image lumpectomy margins surrounding a mass in 48 patients 
. What is unique about our published work on breast tumor margin assessment is that we demonstrated the capability to image an entire tumor margin, which has yet to be demonstrated by previously published optical techniques. The engine of this bench-top spectral imaging system is a broadband source that emits at visible wavelengths, an imaging spectrograph, and a CCD camera which are shown in
. Light is relayed between the instrument and each discrete site on the margin within a specimen box via an imaging probe () 
. The diffuse reflectance spectra per
site were analyzed with a feature extraction algorithm based on a fast, scalable Monte Carlo model developed by our group 
to quantitatively determine absorption (β-carotene and hemoglobin) and scattering contrast in the breast. These sources of contrast were used to create tissue morphology maps which were used in a decision-tree model to differentiate positive from negative margins. We reported sensitivity and specificity of 79% and 67% respectively on 55 margins from 48 patients 
imaged 16±5 minutes post-excision. We have since accrued images from 88 margins in 70 patients and the results are consistent with those reported previously. In summary, optical imaging technologies can aid the surgeon in finding positive margins and they can also be used to guide pathological assessment of tissue and provide insight into where to sample the tissue, thereby improving sampling yield, particularly in larger tumor specimens in both the intra-operative and post-operative setting.
Instrumentation and measurement procedure.
Before this technology can be used in an intra-operative setting or in a post-operative setting, systematic studies have to be performed to determine which surgical and post-surgical factors affect the precision and accuracy with which this technology maps optical contrast. This is true not only for our technology but other technologies, both optical and non-optical that are intended for this application. Specifically, if the technology is to be used on the excised margin (which is the way in which intra-operative pathology is performed), then there must be an understanding of how the presence of the blue sentinel lymph node mapping dye (referred to as patent blue dye) and cautery could influence the primary sources of contrast in the breast. Another important variable to characterize is the impact of the time delay after excision on the primary sources of optical contrast in the breast. Given that all of the recent studies reporting on optical technologies have been carried out on resected tumor margins 
and the fact that frozen section and post-operative pathology are necessarily carried out on resected specimens, characterizing the effects of these potential sources of error will be important in the context of developing optically based margin imaging tools for use by surgeons and pathologists.