Basal cell carcinoma (BCC) constitutes about 75% of skin cancers, with more than 1.7 million new patients diagnosed each year in the United States and 60,000 in the United Kingdom. For aggressive BCCs, Mohs micrographic surgery is considered the most suitable treatment. In this type of surgery, one layer of tissue after another is cut away and examined under the microscope to make sure that all the cancer is out. This process is stopped when only healthy tissue is left. There is always a balance to be struck between making sure that all the cancer is removed and preserving as much healthy tissue as possible in order to reduce scarring and disfigurement. The main disadvantage of Mohs surgery is the need for frozen section preparation and histopathology examination for all excised tissues, a non-automated, time-consuming technique (1–2 hours per layer).
Techniques based on optical microscopy are very attractive for tissue diagnosis as they allow the observation of individual cells and tumors with micron-scale spatial resolution. In particular, imaging based on molecular spectroscopy has the advantage that it contains both morphological and chemical tissue information. Previous work showed that Raman microscopy can diagnose and map BCC regions with accuracy similar to gold-standard histopathology.1,2 Nevertheless, the raster scanning used to build spectral images required long data acquisition and analysis times (5 hours per 1mm2), making it impractical for use during surgery.
To reduce the data acquisition time and make Raman spectroscopy suitable for intra-operative use, we have developed a selective sampling technique based on multimodal spectral imaging. We first used tissue autofluorescence imaging, which has high sensitivity, high speed, and low specificity, to determine the main spatial features of the sample with high spatial resolution. This information was then used in an automated manner to select and prioritize the sampling points for Raman spectroscopy.3
We developed a classification model for tissue structures based on the Raman spectra. Raman spectroscopy was able to detect molecular differences between tumors and other healthy tissue structures with sensitivity and specificity higher than 94% (see Figure 1). These molecular differences were mainly attributed to increased nucleic acids and decreased collagen in the tumor regions compared to healthy tissue.3
Figure 1. Typical examples of basal cell carcinoma (BCC) by raster scanning Raman micro-spectroscopy (RMS). Scale bar: 400 microns. H&E: Hematoxylin and eosin staining. Inflamed D: Inflamed dermis.
To obtain fast diagnosis of large skin tissue sections (∼1cm2), we used automated segmentation of autofluorescence images to select the sampling points for Raman spectroscopy, which was then used to establish the diagnosis based on a spectral classification model. This automated sampling strategy allowed objective diagnosis of basal cell carcinoma in skin tissue samples based on only 500–1500 Raman spectra (20–60 minutes), which is faster than frozen section histopathology and more than two orders of magnitude faster than previous techniques based on raster scanning Raman microscopy.
Furthermore, Raman spectroscopy relies on inelastic scattering of light by molecules in tissue rather than absorption. Therefore, Raman spectra can be obtained without cutting the tissue into thin sections. Using segmented images from confocal autofluorescence imaging, we demonstrated accurate diagnosis for thick layers of skin obtained directly from surgery with no tissue preparation (see Figure 2).3
Figure 2. Typical example of BCC in a unsectioned skin layer obtained during Mohs surgery. Diagnosis was obtained with 600 Raman spectra. Scale bar: 2mm. MMS: Mohs micrographic surgery. MSH: Multimodal spectral histopathology.
In conclusion, automated imaging and objective diagnosis of tissue specimens during cancer surgery is a promising approach for improving this advanced surgical procedure. This quantitative approach can increase the efficacy of surgery by eliminating the errors related to the subjective inter-observer evaluation of histopathological sections. In addition, the multimodal spectral imaging can be applied to tissue sections as well as tissue block, thus eliminating the time-consuming procedures required to prepare frozen sections for histopathology.
While this study focused on BCC, multimodal spectral imaging is a platform technology and may be used to provide intra-operative diagnosis and ensure clear margins during surgery for other cancer types, for which histopathological diagnosis is not routinely performed at present. Our focus now is to develop an optimized prototype instrument that can be tested in the clinic.
This paper presents independent research commissioned by the National Institute for Health Research (NIHR) under its Invention for Innovation (i4i) Programme (grant II-AR-0209-10012). The views expressed are those of the author(s) and not necessarily those of the National Health Service, the NIHR, or the Department of Health.
University of Nottingham
Ioan Notingher, associate professor in the School of Physics and Astronomy, focuses on biomedical applications of Raman micro-spectroscopy ranging from characterization of bio-nanomaterials, noninvasive imaging of live cells, and intra-operative imaging of tumors for cancer surgery.
1. A. Nijssen, T. C. Bakker Schut, F. Heule, P. J. Caspers, D. P. Hayes, M. H. Neumann, G. J. Puppels, Discriminating basal cell carcinoma from its surrounding tissue by Raman spectroscopy, J. Invest. Dermatol. 119, p. 64-69, 2002.
2. M. Larraona-Puy, A. Ghita, A. Zoladek, W. Perkins, S. Varma, I. H. Leach, A. A. Koloydenko, H. Williams, I. Notingher, Development of Raman microspectroscopy for automated detection and imaging of basal cell carcinoma, J. Biomed. Opt.
14, p. 054031, 2009. doi:10.1117/1.3251053
3. K. Kong, C. J. Rowlands, S. Varma, W. Perkins, I. H. Leach, A. A. Kolydenko, H. C. Williams, I. Notingher, Diagnosis of tumors during tissue-conserving surgery with integrated autofluorescence and Raman scattering microscopy, Proc. Nat'l Acad. Sci. USA 110(38), p. 15189-15194, 2013.