Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Multimodality microscopy and micro-Raman spectroscopy for in vivo skin characterization and diagnosis Wang, Hequn 2013

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
24-ubc_2013_spring_wang_hequn.pdf [ 17.29MB ]
Metadata
JSON: 24-1.0073699.json
JSON-LD: 24-1.0073699-ld.json
RDF/XML (Pretty): 24-1.0073699-rdf.xml
RDF/JSON: 24-1.0073699-rdf.json
Turtle: 24-1.0073699-turtle.txt
N-Triples: 24-1.0073699-rdf-ntriples.txt
Original Record: 24-1.0073699-source.json
Full Text
24-1.0073699-fulltext.txt
Citation
24-1.0073699.ris

Full Text

Multimodality Microscopy and Micro-Raman Spectroscopy for In Vivo Skin Characterization and Diagnosis by Hequn Wang B.Eng., Huazhong University of Science and Technology, 2007  A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF  Doctor of Philosophy in THE FACULTY OF GRADUATE STUDIES (Interdisciplinary Oncology)  The University Of British Columbia (Vancouver) April 2013 c Hequn Wang, 2013  Abstract  Accurate and early diagnosis of skin diseases will improve clinical outcomes. Visual inspection alone has limited diagnostic accuracy, while biopsy followed by histopathology examination is invasive and time-consuming. The objective is to design and develop a multimodal optical instrument that provides biochemical and morphological information on human skin in vivo. Raman spectroscopy (RS) is capable of providing biochemical information of tissues. Reflectance confocal microscopy (RCM), which generates micron-level resolution images with capability of optical sectioning, can provide refractive-index-based morphological information of the skin. Multiphoton microscopy (MPM) could simultaneously provide biochemistry-based morphological information from two-photon fluorescence (TPF) and second-harmonic-generation (SHG) images. The thesis hypothesis is that a multimodality instrument combining RS, RCM, and MPM could be developed and provide complementary information in real-time for in vivo skin evaluation and aiding non-invasive diagnosis. A confocal Raman spectroscopy system was initially developed and tested in a study on in vivo mouse skin. Spectral biomarkers (899 and 1325-1330 cm −1 ) were found to differentiate tumor-bearing skin from normal skin. A RCM system was then integrated with the spectroscopy system to guide spectral measurements. Noninvasive morphological and biochemical analysis was performed on ex vivo and in vivo human skin. The system was further enhanced by adding an MPM module that can image cellular structures with TPF signals from keratin, NADH, and melanin, as well as image elastic and collaii  gen fibers via TPF and SHG signals, respectively. The finalized system was utilized to noninvasively measure a cherry angioma lesion and its surrounding structures on the skin of a volunteer. Confocal Raman spectra from various regions-of-interest acquired under the guidance of MPM and RCM imaging showed different spectral patterns for blood vessels, keratinocytes, and dermal fibers. The system was also successfully used to perform imaging directed two-photon absorption based photothermolysis on ex vivo mouse skin. All the results showed positive evidence, well supporting the overall hypothesis. The developed multimodality system, capable of acquiring co-registered RCM, TPF and SHG images simultaneously at video-rate, and performing image-guided region-of-interest Raman spectral measurements of human skin in vivo, is a powerful tool for non-invasive skin evaluation and diagnosis.  iii  Preface Partial contents of this dissertation have been published or submitted for publication. The details can be found below. A version of Sections 2.2 and 2.4 in Chapter 2 has been published as: Wang, H., Huang, N., Zhao, J., Lui, H., Korbelik, M. and Zeng, H. (2011), Depth-resolved in vivo micro-Raman spectroscopy of a murine skin tumor model reveals cancer-specific spectral biomarkers. Journal of Raman Spectroscopy, 42: 160-166. For this publication, I partially designed the study, constructed the system, performed all the experiments, analyzed the data, and prepared the manuscript. Huang, N. helped in handling the animals and Zhao, J. helped in data analysis; Lui, H. and Zeng, H. helped in the study design and provided feedbacks to the manuscript. Korbelik, M. provided the animals and edited the manuscript. A version of Section 2.5 in Chapter 2 has been published as: Wang, H., Zhao, J., Lee, A.M.D., Lui, H., and Zeng, H. (2012), Improving skin Raman spectral quality by fluorescence photobleaching. Photodiagnosis and Photodynamic Therapy, 9: 299-302. For this publication, I partially identified the project and designed the study, performed all the experiments, analyzed the data, and prepared the manuscript. Zhao, J. constructed the macro-Raman spectroscopy system and helped in data analysis; Lee, A.M.D. provided essential comments to the manuscript. Lui, H., and Zeng, H. helped in identifying the project and edited the manuscript. A version of Sections 4.1, 4.2, and 4.5 in Chapter 4 has been published as: Lee, A.M.D, Wang, H., Yu, Y., Tang, S., Zhao, J., Lui, H., McLean, D.I., Zeng, H. (2011), In vivo  iv  video rate multiphoton microscopy imaging of human skin. Optics Letters, 36(15): 28652867. Lee, A.M.D and myself are co-first authors and had equal contributions to this publication. For this publication,I partially designed and built the system, performed all the measurements, analyzed all the data, prepared figures and movies for the manuscript. Lee, A.M.D. helped in the system design and development, and wrote the manuscript. Yu, Y. helped in developing the software of the system. Zeng, H, Lui, H, and McLean, D.I. helped in the study design and edited the manuscript. Tang, S. and Zhao, J. provided essential comments to the manuscript. A version of Sections 5.1, 5.2 and 5.5 in Chapter 5 has been published as: Wang, H., Lee, A.M.D., Frehlick, Z., Lui, H., McLean, D.I., Tang, S., and Zeng, H. (2013), Perfectly registered multiphoton and reflectance confocal video rate imaging of in vivo human skin. Journal of Biophotonics, 6(4): 305-309. For this publication, I partially designed the study, developed the system, performed all the measurements, analyzed the data, prepared all the figures and movies, and wrote the manuscript. Lee, A.M.D. helped in the system design and development, and provided essential feedbacks to the manuscript. Frehlick, Z. helped in developing the software of the system. Zeng, H., Lui, H., and McLean, D.I. helped in study design and edited the manuscript. Tang, S. provided essential comments to the manuscript. A version of Chapter 6 has been submitted for publication as: Wang, H., Lee, A.M.D., Lui, H., McLean, D.I., and Zeng, H. (2013), A method for accurate in vivo micro-Raman spectroscopic measurements under guidance of advanced microscopy imaging. Scientific Reports. For this publication, I designed the study, partially constructed the system, performed all the measurements, analyzed the data, prepared all the figures and movies, and wrote the manuscript. Lee, A.M.D. helped in developing the system and provided feedbacks to the manuscript. Zeng, H., Lui, H., and McLean, D.I. helped in study design and edited the manuscript. A version of Chapter 7 has been published online as: Wang, H., Zandi, S., Lee, A.M.D.,  v  Zhao, J., Lui, H., McLean, D.I., and Zeng, H. (2013), Imaging directed photothermolysis through two-photon absorption demonstrated on mouse skin − a potential novel tool for highly targeted skin treatment. Journal of Biophotonics, DOI: 10.1002/jbio.201300016. For this publication, I partially designed the study, constructed the system, performed all the measurements, prepared all the figures and movies, and wrote the manuscript. Zandi, S. helped in the study design and the experiments, and edited the manuscript. Lee, A.M.D. helped in developing the system and provided feedbacks to the manuscript. Zeng, H., Lui, H. and McLean, D.I. helped in study design and edited the manuscript. Zhao, J. provided valuable discussions and comments to the manuscript. All the published contents including figures, tables, and texts, are used with permission. Animal care and experiments were carried out in accordance with the guidelines of the Canadian Council of Animal Care (CCAC), and the use of animals for our experiments was reviewed and approved by the Animal Care Committee of the University of British Columbia (certificate #: A08 − 0577 and A10 − 0338). All the experiments on human volunteers were approved by the University of British Columbia Research Ethics Board (certificate #: H96 − 70499 and H05 − 70569).  vi  Table of Contents Abstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .  ii  Preface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .  iv  Table of Contents . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . vii List of Tables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .  xi  List of Figures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xii List of Abbreviations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xvii Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xxi Dedication . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xxiii 1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .  1  1.1  Structure and function of human skin . . . . . . . . . . . . . . . . . . . . .  1  1.2  Common diseases of human skin . . . . . . . . . . . . . . . . . . . . . . .  3  1.3  Overview of noninvasive optical techniques used in dermatology and skin research . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .  5  1.3.1  Wood’s lamp . . . . . . . . . . . . . . . . . . . . . . . . . . . . .  5  1.3.2  Dermoscopy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .  6  1.3.3  Spectroscopy . . . . . . . . . . . . . . . . . . . . . . . . . . . . .  6  vii  1.3.4 1.4  Advanced optical imaging . . . . . . . . . . . . . . . . . . . . . . 14  Objective and outline of this dissertation . . . . . . . . . . . . . . . . . . . 22  2 In Vivo Confocal Raman Spectroscopy . . . . . . . . . . . . . . . . . . . . . . 25 2.1  Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25  2.2  System design and development . . . . . . . . . . . . . . . . . . . . . . . 27  2.3  Resolution measurement and system calibration . . . . . . . . . . . . . . . 28  2.4  Measurements on in vivo mouse skin . . . . . . . . . . . . . . . . . . . . . 31  2.5  2.6  2.4.1  Data processing and statistical analysis . . . . . . . . . . . . . . . . 32  2.4.2  Classification of normal mouse skin and malignant mouse skin . . . 34  2.4.3  Discussion and conclusion . . . . . . . . . . . . . . . . . . . . . . 41  Raman spectral quality improvement . . . . . . . . . . . . . . . . . . . . . 44 2.5.1  Materials and methods . . . . . . . . . . . . . . . . . . . . . . . . 45  2.5.2  Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47  2.5.3  Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51  Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52  3 In Vivo Confocal Raman Spectroscopy with Imaging Guidance . . . . . . . . 54 3.1  Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54  3.2  System design and development . . . . . . . . . . . . . . . . . . . . . . . 55  3.3  Measurements on ex vivo skin samples . . . . . . . . . . . . . . . . . . . . 57  3.4  Measurements on in vivo human skin . . . . . . . . . . . . . . . . . . . . . 58  3.5  Discussion and conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . 60  4 In Vivo Video Rate Multiphoton Microscopy . . . . . . . . . . . . . . . . . . 63 4.1  Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63  4.2  System design and development . . . . . . . . . . . . . . . . . . . . . . . 65  4.3  Imaging extracted biological samples . . . . . . . . . . . . . . . . . . . . . 67  4.4  Imaging ex vivo human skin . . . . . . . . . . . . . . . . . . . . . . . . . . 67 viii  4.5  Imaging in vivo human skin . . . . . . . . . . . . . . . . . . . . . . . . . . 67  4.6  Volumetric imaging demonstrating optical biopsy . . . . . . . . . . . . . . 71 4.6.1  Volumetric imaging on extracted biological samples . . . . . . . . . 74  4.6.2  Volumetric imaging on ex vivo human skin . . . . . . . . . . . . . . 74  4.6.3  Volumetric imaging on in vivo human skin . . . . . . . . . . . . . . 76  4.7  Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78  4.8  Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80  5 Co-registered Multiphoton and Reflectance Confocal Video Rate Imaging of In Vivo Human Skin . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81 5.1  Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81  5.2  System design and development . . . . . . . . . . . . . . . . . . . . . . . 84  5.3  Imaging extracted biological samples . . . . . . . . . . . . . . . . . . . . . 86  5.4  Imaging ex vivo human skin . . . . . . . . . . . . . . . . . . . . . . . . . . 87  5.5  Imaging in vivo human skin . . . . . . . . . . . . . . . . . . . . . . . . . . 88  5.6  Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 93  5.7  Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 94  6 Raman Spectroscopy Combined with Reflectance Confocal Imaging and Multiphoton Microscopy Imaging System . . . . . . . . . . . . . . . . . . . . . . 96 6.1  Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 96  6.2  System design and development . . . . . . . . . . . . . . . . . . . . . . . 98  6.3  Measurements on in vivo human skin . . . . . . . . . . . . . . . . . . . . . 99  6.4  Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105  6.5  Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 108  7 Imaging Directed Photothermolysis through Two Photon Absorption . . . . . 109 7.1  Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 109  7.2  Materials and methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . 111 ix  7.2.1  Animal preparation . . . . . . . . . . . . . . . . . . . . . . . . . . 111  7.2.2  Integrated reflectance confocal microscopy and two-photon imaging system . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 111  7.3  Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 116  7.4  Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 123  7.5  Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 125  8 Conclusion and Future Directions . . . . . . . . . . . . . . . . . . . . . . . . 127 8.1  Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 127 8.1.1  Confocal Raman spectroscopy . . . . . . . . . . . . . . . . . . . . 127  8.1.2  Confocal Raman spectroscopy integrated with reflectance confocal microscopy imaging . . . . . . . . . . . . . . . . . . . . . . . . . 128  8.2  8.1.3  Video rate multiphoton and reflectance confocal microscopy . . . . 128  8.1.4  Multimodality microscopy and spectroscopy . . . . . . . . . . . . . 129  8.1.5  Two-photon absorption based photothermolysis . . . . . . . . . . . 130  Future directions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 131  Bibliography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 134 Appendix . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 163  x  List of Tables Table 2.1  Raman peak assignment for ex vivo mouse skin . . . . . . . . . . . . . . 38  Table 6.1  Raman peak assignment for in vivo human skin with a cherry angioma lesion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 103  Table 7.1  Combinations of target depth and laser exposure duration. . . . . . . . . 115  xi  List of Figures Figure 1.1  Diagram of human skin. . . . . . . . . . . . . . . . . . . . . . . . . .  2  Figure 1.2  H&E (hematoxylin and eosin) stain of human skin. . . . . . . . . . . .  3  Figure 1.3  Physical processes of Rayleigh scattering, Stokes Raman scattering, and Anti-Stokes Raman scattering. . . . . . . . . . . . . . . . . . . . .  9  Figure 1.4  Example Raman spectrum of ethanol. . . . . . . . . . . . . . . . . . . 10  Figure 1.5  Configuration of a reflectance confocal microscope. . . . . . . . . . . . 16  Figure 1.6  Diagram of some physical processes relevant to two-photon fluorescence (TPF) and second harmonic generation (SHG). . . . . . . . . . . 20  Figure 2.1  Schematic configuration of the confocal Raman spectrometer system for depth-resolved skin Raman measurements. . . . . . . . . . . . . . . 28  Figure 2.2  Integrated Raman intensity versus lateral displacement on a silicon wafer. 29  Figure 2.3  Integrated Raman intensity versus axial displacement along a silicon wafer. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30  Figure 2.4  Fluorescence background removal of an example Raman spectrum from mouse skin. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32  Figure 2.5  Photomicrographs of mouse skin from normal skin and tumor (H&E). . 35  Figure 2.6  Mean normalized spectra for normal and tumor site skin at different depths from 24 mice. . . . . . . . . . . . . . . . . . . . . . . . . . . . 36  xii  Figure 2.7  Comparison of mean normalized spectra between normal skin and tumor site skin at the epidermis and dermis. . . . . . . . . . . . . . . . . 37  Figure 2.8  Four PCs from PCA analysis of spectra at the epidermal level. . . . . . 38  Figure 2.9  Three PCs from PCA analysis of spectra at the dermal levels. . . . . . . 39  Figure 2.10 Comparison of mean normalized spectra for normal and tumor site skin at the epidermal level with standard deviation. . . . . . . . . . . . . . . 39 Figure 2.11 Scatter plot of three PC scores (PC1, PC2, and PC3) at the dermal levels. 40 Figure 2.12 ROC curve generated from the PCA-LDA analysis of the dermal spectra using the three PCs (AUC = 0.96). . . . . . . . . . . . . . . . . . . 40 Figure 2.13 Ex vivo mouse skin fluorescence decay curves with different scanning areas and at different depths. . . . . . . . . . . . . . . . . . . . . . . . 48 Figure 2.14 Example data from a female volunteer inner forearm in vivo measurements with 1 s detector integration time.  . . . . . . . . . . . . . . . . 50  Figure 2.15 In vivo human volunteer forehead skin fluorescence and Raman measurements taken at 1 s and at 200 s after laser exposure with 1 s detector integration time. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51 Figure 2.16 In vivo human volunteer forehead skin Raman measurements taken with 0.5 s and 2 s detector integration time with 5-point smoothing results of the spectral curves. . . . . . . . . . . . . . . . . . . . . . . . 52 Figure 3.1  Schematic configuration of the system for confocal imaging guided skin Raman measurements. . . . . . . . . . . . . . . . . . . . . . . . . 56  Figure 3.2  Raman spectra and reflectance confocal images of ex vivo human skin. . 57  Figure 3.3  Raman spectra and reflectance confocal images of in vivo human skin. . 59  Figure 3.4  Raman spectra and reflectance confocal images of in vivo human skin (with scanning). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60  Figure 4.1  In vivo video rate MPM setup. . . . . . . . . . . . . . . . . . . . . . . 66  xiii  Figure 4.2  Multiphoton fluorescence and second-harmonic-generation images of extracted bovine elastin and bovine collagen. . . . . . . . . . . . . . . 68  Figure 4.3  Two photon fluorescence (TPF) and second-harmonic-generation (SHG) images of an ex vivo human skin sample. . . . . . . . . . . . . . . . . . 68  Figure 4.4  TPF images from epidermis (oblique orientation) extracted from an in vivo video from the dorsal forearm of a 25-year-old Asian female increasing in depth from near the surface to near the dermal/epidermal boundary. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71  Figure 4.5  Integrated SHG+TPF images of the epidermal ridges and the papillary dermis extracted from an in vivo video of the dorsal forearm of a 28year-old Asian male. . . . . . . . . . . . . . . . . . . . . . . . . . . . 72  Figure 4.6  False color overlay of SHG (green) and TPF (red) images from the reticular dermis extracted from an in vivo video of the dorsal forearm of a 63-year-old Caucasian male. . . . . . . . . . . . . . . . . . . . . . 73  Figure 4.7  Illustration of stacks of images in XY-plane and XZ-plane. . . . . . . . 73  Figure 4.8  3D reconstruction of collagen fibers from XY-Z scan on left and XZ-Y scan on right. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74  Figure 4.9  3D reconstruction of ex vivo human cheek skin from XY-Z scan. . . . . 75  Figure 4.10 3D reconstruction of ex vivo human eyelid skin from XY-Z scan. . . . . 76 Figure 4.11 3D reconstruction of in vivo human forearm skin from XY-Z scan. . . . 77 Figure 4.12 3D reconstruction of in vivo human sweat duct from XY-Z scan. . . . . 77 Figure 5.1  In vivo video rate RCM/MPM system setup. The SHG/TPF dichroic and filters preceding PMTs were changed or removed according to the desired imaging modalities. . . . . . . . . . . . . . . . . . . . . . . . . 85  Figure 5.2  (RCM, SHG+TPF, and false color overlay of SHG+TPF (green) and RCM (red) images from extracted bovine collagen sample. . . . . . . . 86  xiv  Figure 5.3  RCM, SHG+TPF, and false color overlay of SHG+TPF (green) and RCM (red) images from extracted bovine elastin sample. . . . . . . . . 86  Figure 5.4  RCM, SHG+TPF, and false color overlay of SHG+TPF (green) and RCM (red) images from epidermis of ex vivo eyebrow skin from surgery. 87  Figure 5.5  RCM, SHG+TPF, and false color overlay of SHG+TPF (green) and RCM (red) images from dermis of ex vivo eyelid skin from surgery. . . 88  Figure 5.6  RCM, SHG+TPF, false color overlay of SHG+TPF (green) and RCM (red), and false color overlay of SHG+TPF (red) and RCM (green) images extracted from Supplementary Media 9 from the stratum basale (SB) of the dorsal forearm of a Caucasian male in his early 50’s. . . . . 89  Figure 5.7  RCM, SHG+TPF, and false color overlay of SHG+TPF (green) and RCM (red) images extracted from Supplementary Media 10 from the stratum granulosum (SG) of the dorsal forearm of a 41-year-old Asian male. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 90  Figure 5.8  RCM, TPF, SHG, and false color overlay of RCM (red), TPF (green), and SHG (blue) images extracted from Supplementary Media 11 from the reticular dermis (RD) of the ventral forearm of a 64-year-old Caucasian male.  Figure 5.9  . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91  RCM, SHG+TPF, false color overlay of SHG+TPF (green) and RCM (red), and false color overlay of SHG+TPF (red) and RCM (green) images from the dorsal forearm of a 41-year-old Asian male volunteer. . . 92  Figure 6.1  Diagram of optical prism compressor. . . . . . . . . . . . . . . . . . . 99  Figure 6.2  System diagram of in vivo multimodal confocal Raman spectroscopy, reflectance confocal and multiphoton imaging. . . . . . . . . . . . . . . 100  Figure 6.3  In vivo integrated SHG+TPF images extracted from Supplementary Media 12 of the dorsal forearm of a 23-year-old Asian male (optic compressor introduced into the system). . . . . . . . . . . . . . . . . . . . 101 xv  Figure 6.4  In vivo confocal Raman spectra of blood in a cherry angioma acquired under guidance of reflectance confocal imaging and multiphoton imaging.102  Figure 6.5  In vivo confocal Raman spectra of fibers and cells surrounding a cherry angioma acquired under guidance of reflectance confocal imaging and multiphoton imaging. . . . . . . . . . . . . . . . . . . . . . . . . . . . 104  Figure 6.6  Four consecutive video frames used to derive blood velocity. . . . . . . 105  Figure 6.7  In vivo monitoring of changes of blood glucose level of a cherry angioma lesion on the upper arm skin of a volunteer using confocal Raman spectroscopy guided with reflectance confocal imaging. . . . . . . 106  Figure 7.1  Diagram demonstrating the horizontal view (X-Y plane) of the epidermal cuts, the line of tissue damages at dermis layer, and the histological sectioning. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 114  Figure 7.2  RCM and SHG+TPF images comparing before, during and after high power fs laser irradiation. . . . . . . . . . . . . . . . . . . . . . . . . . 116  Figure 7.3  H&E histological photomicrographs demonstrating selective alteration in laser-exposed mouse skin. . . . . . . . . . . . . . . . . . . . . . . . 118  Figure 7.4  SHG+TPF images of different depths comparing before and after high power (200 mW ) laser irradiation with a target treatment depth of 40  μ m and a irradiation period of 2 min. . . . . . . . . . . . . . . . . . . . 119 Figure 7.5  SHG+TPF images of different depths comparing before and after high power (200 mW ) irradiation with a target depth of 30 μ m and an irradiation period of 2 min. . . . . . . . . . . . . . . . . . . . . . . . . . . 120  Figure 7.6  Comparison of CW laser-induced tissue alteration and fs laser-induced tissue alteration. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 122  Figure 7.7  Proposed explanation for tissue elevation during tissue alteration. . . . . 124  xvi  List of Abbreviations APD  Avalanche photodiode  AUC  Area under curve  BCC  Basal cell carcinoma  BS  Beamsplitter  CARS  Coherent anti-Stokes Raman scattering  CCD  Charge coupled device  CSLM  Confocal scanning laser microscopy  CW  Continuous wave  cwRCM Continuous wave reflectance confocal microscopy DAQ  Data acquisition  DEJ  Dermal-epidermal junction  DRS  Diffuse reflectance spectroscopy  FCM  Fluorescence confocal microscopy  FOV  Field of view  f ps  Frames per second xvii  fs  Femtosecond  f sRCM Femtosecond reflectance confocal microscopy FT  Fourier transform  FWHM  Full width at the half maximum  H&E  Hematoxylin and eosin  Hb  Hemoglobin  H . Sync  Horizontal synchronization  HbO2  Oxyhemoglobin  IR  Infrared  LDA  Linear discriminant analysis  LP  Long pass  MPM  Multiphoton microscopy  N.A.  Numerical aperture  NAD  Nicotinamide adenine dinucleotide  NAD ( P ) H  Reduced nicotinamide adenine dinucleotide (phosphate)  NIR  Near infrared  NMF  Natural moisturizing factor  NMSC  Non-melanoma skin cancer  OCT  Optical coherence tomography  PBS  Polarization beamsplitter xviii  PC  Principal component  PCA  Principal component analysis  PC - GDA  Principal component with generalized discriminant analysis  PD  Papillary dermis  PLS  Partial least-squares  PMT  Photomultiplier tube  RCM  Reflectance confocal microscopy  RD  Reticular dermis  ROC  Receiver operating characteristic  ROI  Region of interest  RS  Raman spectroscopy  SAAID  SHG to autofluorecence aging index of dermis  SERS  Surface-enhanced Raman scattering  SB  Stratum basale  SC  Stratum corneum  SCC  Squamous cell carcinoma  SG  Stratum granulosum  SHG  Second harmonic generation  SNR  Signal-to-noise ratio  SPRR  Small proline-rich protein xix  SS  Stratum spinosum  TPA  Two-photon absorption  TPF  Two-photon fluorescence  UV  Ultraviolet  V. Sync  Vertical synchronization  W. D.  Working distance  xx  Acknowledgments First of all, I would like to sincerely acknowledge my dear supervisor Dr. Haishan Zeng whose guidance and supervision made all this work possible. All kinds of your brilliant ideas, your great cares and understanding to the students, your positive attitude to life, and your profound and thorough perspectives will influence me and be memorized through my whole life. And I hope to let you know that you are a great supervisor and I am really lucky to be one of your students. I learnt from you more than anyone can think of, and there is no such a word to express my gratitude. Secondly, I would like to say thank you to my committee members Drs. Calum MacAulay, Harvey Lui, and David McLean for their helpful suggestions on my work. Special thanks to Drs. Lui and McLean for their great advices and guidance throughout my whole PhD study. You two taught me not only how to be a good student, but also how to be a good researcher, a good mentor. You are my great role models. Special thanks to Dr. Anthony Lee for teaching me so many basic principles in optics, useful skills on optical alignment, and for providing me endless and substantial thoughts and helps over many projects. Also, I thank Dr. Jianhua Zhao and Dr. Michael Short for sharing their valuable experience in Raman spectroscopy, Mr. Yingqiu Yu, Mr. Zack Frehlick, Ms. Irene Tong for providing great software support, Mr. Wei Zhang, Mr. Soroush Merchant, and Dr. Soodabeh Zandi for their assistance in the animal experiment. Without your helps and support, I would never be able to complete my study. I appreciate all the group members from Dr. Haishan Zeng’s lab and from the Pho-  xxi  tomedicine Institute in the Department of Dermatology and Skin Science at University of British Columbia, your suggestions and encouragements made me feel very supportive, just like in a big family. You have no idea how lucky and how happy I feel to be one of our group members. I am also very grateful for all the helps from the colleagues and friends in the Integrative Oncology Department at BC Cancer Research Center. Special thanks to Dr. Gerald Li, Ms. Hanna Pawluk, and Mr. Kam Chow for their valuable suggestions to my dissertation. Thank you all!  xxii  Dedication I would like to thank my parents, my husband and all my family members for their selfgiving support and love. All my accomplishments would not have happened without you, thank you very much!  xxiii  Chapter 1 Introduction 1.1 Structure and function of human skin The skin is the largest organ of the human body. It consists of a stratified, cellular epidermis and underlying dermis of connective tissue [1] (shown in Figure 1.1). The dermalepidermal junction (DEJ), which is undulating, provides mechanical support for the epidermis and is regarded as a partial barrier against exchange of cells and large molecules. Under the dermis is the subcutaneous fat layer. The skin epidermis can be divided into sublayers: stratum corneum (SC), which is the most superficial layer containing non-viable keratinocytes and serves as the skin’s principal permeability barrier layer; stratum granulosum (SG), which is dominated by granular cells promoting hydration and crosslinking of keratin; stratum spinosum (SS), which contains spinous cell and is the place where keratinization process begins; and stratum basale (SB), which provides germinal cells necessary for the regeneration of the layers of the epidermis and is the deepest of the four layers of epidermis. A photomicrograph of normal human skin is shown in Figure 1.2, where different sub-layers of skin epidermis can be seen. The dermis, which consists of a large amount of collagen and elastic fibers embedded in an amorphous ground substance material, can also be divided into two sub-layers: papillary dermis (PD), which is adjacent to the epidermis and is composed of dermal papilla and loosely arranged collagen and elastic fibers; 1  stratum granulosum stratum spinosum stratum basale  Subcuteneous fat  Figure 1.1: Diagram of human skin. (Source: http://commons.wikimedia.org) reticular dermis (RD), which is under the layer of papillary dermis and composed of densely packed collagen and elastic fibers [2]. The dermis also contains a variety of appendages of the skin, such as hair follicles, arrector pili muscles, sweat glands, and sebaceous glands. The subcutaneous fat layer, the lowermost layer of the integumentary system in humans, mainly contains adipose cells, fibroblasts, and macrophages. It stores energy, protects and cushions blood vessels and nerve fibers, and also aids in the process of homeostasis by forming a layer of insulation to slow heat loss. The main functions of skin include sensation, protection, thermal regulation, and secretion. To be more specific, temperature and pressure receptors allow us to react to external  2  stratum corneum  stratum granulosum stratum spinosum  stratum basale  papillary dermis reticular dermis  Figure 1.2: H&E (hematoxylin and eosin) stain of human skin. http://commons.wikimedia.org)  (Source:  stimuli and to interpret the outside environment. Melanin produced by the melanocytes in the epidermis protects us from the ultraviolet (UV) rays of the sunlight. Blood vessels in the dermis will dilate to bring more blood flow to the surface of the body when there is an external temperature increase. Moreover, sebum and sweat are two important secretions produced by the skin.  1.2 Common diseases of human skin There are numerous skin conditions that affect people in their daily life. Some skin conditions are minor or even barely visible. However, there are some skin diseases that are serious and potentially life-threatening, such as skin cancer [3]. Common skin conditions include acne, seborrheic keratosis, dermatitis, infections, psoriasis and skin cancer. Skin cancer affects a large proportion of the population in the world. Skin cancer is also the most common type of cancer in Canada and the United States [4, 5], and one  3  in five North Americans are expected to develop skin cancers during their lifetimes [6]. Common skin cancers can be subdivided into two major categories: melanoma (malignant melanoma) and nonmelanoma skin cancers (NMSCs) [7]. Common NMSCs include basal cell carcinoma (BCC) and squamous cell carcinoma (SCC). Nonmelanoma skin cancer has relatively a low rate of metastasis. Malignant melanoma accounts for less than 5% of all skin cancer cases; however, it represents the vast majority of skin cancer deaths [5]. An estimation of 81, 300 new cases and 320 deaths from NMSC is expected in Canada in 2012 [4]. For melanoma, the number of estimated new cases is 5, 800 with estimated 970 deaths. In United States, melanoma is expected to be diagnosed in about 76, 250 persons with an estimated 9, 180 deaths in 2012, and the incidence rates of melanoma have been increasing by almost 3% per year since 2004 [5]. When a suspicious lesion is clinically detected by a physician, biopsy followed by histopathologic examination is the gold standard for confirming the diagnosis. This process is time consuming and can be associated with some morbidity. The importance of achieving high diagnostic sensitivity necessitates a low threshold for biopsy, which in turn incurs higher costs for the healthcare system. Furthermore, a biopsy alters the site under study and leaves a permanent scar on patient’s skin. Finally, the most appropriate site to biopsy is sometimes difficult to determine. Accurate diagnosis of these common skin diseases especially skin cancers is important to determine proper treatment procedures, and will eventually improve the clinical outcome and quality of life of the patients. Diagnosis for most of the skin conditions is solely based on visual inspection by physicians, which is subjective and may limit the diagnosic accuracy. Moreover, there are a significant number of skin conditions that will be diagnosed by either family doctors or general physicians, who are non-dermatologists with a lack of special training in dermatology. This may lead to a higher number of misdiagnoses for many skin conditions. A retrospective biopsy study that correlates the clinical diagnoses of family physicians, plastic, general and orthopedic surgeons, and dermatologists with the  4  histopathologic diagnoses was performed [8], where 4451 cases were analyzed in total. For neoplastic and cystic skin lesions confirmed by histopathological diagnosis, dermatologists diagnosed 75% of the cases correctly, while non-dermatologists only diagnosed 40% of the cases correctly. For inflammatory skin diseases, the diagnosic accuracy from dermatologists and non-dermatologists were 71% and 34%, respectively. Therefore, better diagnostic tools, especially for non-dermatologists, are needed.  1.3 Overview of noninvasive optical techniques used in dermatology and skin research Since dermatology is a medical specialty that is intensively dependent on visualization and pattern recognition, developments and improvements in imaging and spectroscopic technology will be greatly helpful. Moreover, the standard biopsy method is invasive, uncomfortable, and inconvenient; therefore, features such as noninvasive analysis and diagnostic accuracy should be emphasized when developing new technologies or instruments.  1.3.1 Wood’s lamp The idea of using fluorescence light to help visualize skin has been proposed about one century ago with the first invention of Wood’s lamp in 1903 [9]. It emits light at a wavelength of approximately 365 nm in the UV range, which induces skin fluorescence that is visible to clinicians. The Wood’s lamp was first used in clinical dermatology to detect fungal infection of hair in 1925 [10]. More applications in diagnosing pigmented disorders [11], cutaneous infections [12], and porphyrin disorders [13] proliferated after that. For example, Wood’s lamp was used to determine whether melasma is predominated in the epidermis or dermis [11]. Erythrasma, which is caused by bacterial infection, can be better visualized using Wood’s lamp because it shows coral-red fluorescence under Wood’s light examination [12]. Wood’s light examination is also very useful in diagnosing porphyrin related disorders because excess porphyrins in teeth, urine, or blood could be detected [13].  5  However, there are a few limitations of this technique, including: (1) Wood’s lamp examination is most reliable in lighter skin types because a low level of endogenous melanin is preferred for this type of examination to detect subtle pigmentation changes. (2) Many other exogenous components could generate fluorescence under Wood’s lamp examination such as soap residue and topical medications, which may interfere with the tissue autofluorescence.  1.3.2 Dermoscopy Dermoscopy or dermatoscopy, originated from skin surface microscopy, is a noninvasive diagnostic technique for in vivo examination of skin lesions, enabling better visualization of subsurface structures [14]. In 1950s, Goldman was the first dermatologist to use this technique for evaluation of pigmented skin lesions [15]. Previous studies have shown that the diagnostic accuracy of dermoscopy, comparing with clinical visual inspection, increases between 5% and 30%, depending on the type of skin lesion and the experience of the physician [16–19]. Dermoscopy has also been used to image nonpigmented skin tumors [20, 21], inflammatory and infectious diseases [22], and autoimmune diseases [23]. Dermoscopy has also been applied to predict and monitor skin reactions and treatment responses [22, 24, 25]. One of the limitations of dermoscopy is that compared with light microscopy, dermoscopy has a relatively low optical resolution and no sectioning capabilities, resulting in lower diagnostic accuracies for some diseases. For example, when applying dermoscopy for diagnosing melanoma, some studies have revealed that dermoscopy lacks specific dermoscopic criteria, especially for detecting some early melanomas [26–28].  1.3.3 Spectroscopy Since light will be variably absorbed or scattered by different skin components and the skin may re-emit light of different wavelengths, collecting and analyzing the re-emitted light could allow us to interpret the changes of the skin structures and biochemical compositions. Spectroscopy, which studies the emitted radiation intensity as a function of wavelength, 6  can be used to study the interaction between light and skin [29]. Within the broad field of spectroscopy, there are three major techniques that have been used often in dermatology and skin research. Diffuse reflectance spectroscopy The first technique is called diffuse reflectance spectroscopy (DRS), which studies the light absorption and scattering characteristics of the skin. Light is absorbed by a variety of skin chromophores, and scattered due to refractive index differences. DRS has been used to quantitatively assess melanin and hemoglobin inside the skin on in vivo volunteers [30]. Significant differences in the reflectance spectrum were also found between benign nevus and melanoma [31]. One potential limitation with diffuse reflectance spectroscopy is that the information is usually from a relatively large area and the distribution of chromophores is not considered. Therefore, the diagnosis may not be specific and the sensitivity for diagnosing early stage diseases is limited. Fluorescence spectroscopy The second spectral technique that is used in dermatology and skin research is fluorescence spectroscopy. Major skin fluorophores are melanin, hemoglobin, tryptophan, collagen, elastin, nicotinamide adenine dinucleotide (NAD), porphyrins, and flavins [32, 33]. Previous study has shown that endogenous fluorescence of non-melanoma skin cancer has different characteristic features from normal tissue [34]. Fluorescence spectroscopy has also been used to non-invasively study the skin surface thickness of topical agents [35]. However, the optimal excitation wavelength for many fluorophores is within ultraviolet (UV) range, which may cause damage to the skin. The short excitation wavelength will also prevent acquiring signals from deeper skin layers because of the limited penetration depth of UV photons.  7  Raman spectroscopy Raman spectroscopy (RS) is a vibrational spectroscopic technique, which is relatively simple, reproducible, nondestructive and applicable for noninvasive skin measurements [36]. RS is capable of providing details on the chemical composition, molecular structure, and molecular interactions in cells and tissues [37]. When the energy of incident photon is unchanged after interaction with a molecule, the scattered photon has the same frequency as the incident photon. This is referred as Rayleigh or elastic scattering. When energy is transferred either from the photon to molecule or vice versa, the scattered photon has less or more energy than the incident photon and this is referred as inelastic scattering or Raman scattering. Since the molecular energy is quantized, the difference between the incident photon and the scattered photon is also quantized, meaning that the vibrational spectrum is unique to each specific molecule. The Raman effect was first described in 1928 by an Indian physicist Sir C. V. Raman, who received the Nobel Prize in 1930 for his achievement in this field [38]. If the emitted photon has a higher frequency than the incident photon, the process is called anti-Stokes scattering, whereas if the emitted photon has a lower frequency than the incident photon, it is defined as the Stokes shift. A diagram including Rayleigh scattering, Stokes Raman scattering, and anti-Stokes Raman scattering is shown in Figure 1.3. Raman scattering is inherently weak, typically 10−9 to 10−6 of the intensity of the Rayleigh background [39]. Therefore, intense monochromatic excitation and sensitive detection are necessary. A Raman spectrum is a plot of scattered intensity as a function of the frequency difference between the incident and scattered photons and can be obtained by pointing a monochromatic laser beam at a sample [36]. The frequency difference between incident and emitted photons is defined as Raman shift, which is unique for individual molecules and is represented in wavenumbers (1/cm or cm−1 ). The Raman intensity is determined by the concentration and the Raman cross-section of the molecules. The Raman spectrum of an ethanol sample is shown in Figure 1.4. Different Raman peaks represent different 8  Virtual Energy State  Vibrational Energy States  Rayleigh Scattering  Anti-Stokes Stokes Raman Raman Scattering Scattering  Figure 1.3: Physical processes of Rayleigh scattering, Stokes Raman scattering, and Anti-Stokes Raman scattering. chemical bonds of the sample. There are two major types of Raman spectrometers: one is implemented through the use of a dispersive technique, the other by the Fourier transform (FT) technique. In FTRaman spectroscopy, the Raman signals from the sample are divided into two parts. One part is reflected by a beamsplitter and directed onto a fixed mirror. The other part is transmitted and reflected by a moving mirror which has a constant frequency and fixed motion. The two parts re-combine at the beamsplitter and are collected by a detector. Depending on their path difference, the two parts constructively and destructively interfere with each other. Therefore, each data point of the produced interferogram has information about every frequency of the Raman signals collected from the sample. FT-Raman system usually 9  Figure 1.4: Example Raman spectrum of ethanol. Data was kindly provided by Dr. Michael Short from BC Cancer Agency. employs 1064 nm wavelength as excitation, thus has the advantage of reduced fluorescence background from the sample. However, since Raman scattering intensity is proportional to 1/λ 4 , the Raman signals with longer excitation wavelength are weak. On the other hand, a dispersive Raman system usually employs a diffraction grating to disperse different wavelengths and a charge coupled device (CCD) as the detector. Dispersive Raman system has a couple of advantages comparing with FT-Raman. For example, excitation wavelengths can vary from UV, visible, to near-infrared (NIR) ranges. Typical laser wavelengths are 785 nm, 633 nm, 532 nm. Therefore, by changing different wavelengths of excitation, the penetration depth can be controlled, and the autofluorescence from the sample can also be minimized. Compared with FT-Raman system, it also has 10  higher sensitivities due to the shorter excitation wavelengths. The spectral acquisition is much faster than FT-Raman. Moreover, it is easy to be combined with confocal technique to generate a high resolution. Therefore, it is widely used in life science studies. In general, Raman spectrometers usually include 4 basic components: (1) excitation source, which is usually a laser source ranging from ultraviolet (UV) to infrared (IR) and is used to excite the tissue; (2) optics, which may include mirrors, filters, lenses, and fibers to direct and focus laser beams as well as collect Raman signals; (3) monochromator or interferometer (for FT-Raman), which disperses the signals onto a multi-channel detector or is able to simultaneously measure the signal of multiple data points by analysis of signal from a single detector, producing a full spectrum from one scan (FT-Raman); and (4) detector, which is usually a charge coupled device (CCD, a multi-channel detector) capable of discerning a range of wavenumbers simultaneously, or a single channel detector (for FT-Raman), such as photodiode. Raman spectroscopy can be sub-divided into 4 categories based on the scattering types. The most common one is spontaneous RS, which is also called ordinary Raman. The second one is resonance Raman scattering, which may be used to measure low analyte concentrations in a sample by tuning the excitation wavelength to the absorption band of the sample. But the risk of fluorescence and photodegradation of the analyte is increased due to the absorbed energy of incoming laser light. The third one is coherent anti-Stokes Raman scattering (CARS) which is a nonlinear Raman technique achieved when the excitation consists of two overlapping coherent laser beams with a specific frequency difference. CARS needs high peak powers which increase the possibility of sample damage [40]. The fourth one is surface-enhanced Raman scattering (SERS), which enhances Raman scattering by molecules absorbed on rough metal surfaces [41]. SERS involves choosing and/or fabrication of nano-scale structured metal substrates, which introduces more challenges and complexity to the experiment procedure, especially for biological or clinical applications. With the development of lasers and detectors after 1960s, RS has been increasingly  11  used in many applications in medicine and biology [42]. The technique of RS is relatively simple, reproducible, nondestructive to the sample, and with minimal or no sample preparation requirements. Moreover, in biological tissue, RS provides data-rich, fingerprint-type signatures for Raman active molecules, such as collagen, blood, lipids, proteins, and nucleic acids [39, 43]. Therefore, RS has been used in a wide range of biomedical applications, including the study of normal and diseased biological tissues [44, 45], interaction of chemical agents with tissues [46, 47], single cells [48] and implants [49]. As disease leads to changes in the molecular composition of affected tissues, these changes should be recognizable in the spectra. This leads to one of the significant potentials of RS, which is its capability of noninvasive medical diagnosis. Many ex vivo studies in literatures have illustrated that differences can be observed between ex vivo Raman spectra of healthy and diseased tissues, including brain [45], breast [50], cervical tissue [51], skin [52], laryngeal tissue [53], eye [54], colon [55], and artery [56, 57]. To better demonstrate the potential of RS as clinical diagnostic tools, many research groups, including our group, have studied RS extensively over the last decade for its application in vivo [51, 58–61]. A fiber optic probe was designed and used to measure NIR Raman spectra of cervical tissue in vivo [51]. Raman spectra were also acquired from in vivo patients using an NIR fiber optic RS system to demonstrate the feasibility of in vivo RS during clinical routine gastrointestinal endoscopy [58]. RS has also been used in margin assessment during partial mastectomy by collecting in vivo Raman spectra from breast tissue of patients. The potential of RS for reducing the need for re-excision surgeries and recurrence rate of breast cancer has been demonstrated [59]. Blood vessel wall was also measured in vivo by a miniaturized fiber-optic Raman probe, and the results showed that in vivo intravascular Raman signal obtained from a blood vessel is a summation of Raman signal contributions of the blood vessel wall and of the blood [62]. In addition, in vivo Raman spectroscopic measurements were also performed on colon polyps from patients, and a sensitivity of 100% and a specificity of 89% were achieved on differentiating adenomatous polyps from  12  hyperplastic polyps [55]. Our group also developed a NIR endoscopic Raman probe, which was able to acquire Raman spectrum from in vivo lung patients [60]. This demonstrated the potential for improving early detection of lung cancers. Since skin is the largest organ of human being and is easily accessible, Raman spectroscopy has also been successfully applied to study skin. The first Raman spectrum of human skin was published in 1992 [63]. In this study, Fourier transform (FT) Raman was used to study human skin and major Raman peaks of skin were assigned. More studies on the effect of skin hydration, pigmentation, thickness of stratum corenum were carried out [46, 64, 65]. In vivo skin measurements were performed using a rapid NIR RS system developed by our group, which was able to acquire Raman spectrum of skin with less than 1 second integration time and with good signal-to-noise ratio (SNR) [66]. Our group has also undertaken detailed studies of cutaneous melanin [67]. NIR FT-Raman spectroscopy has been used to explore its feasibility for diagnosing basal cell carcinoma, and spectral differences were found in both lipid and protein [68]. Human stratum corneum at different anatomical regions, including fingertip, thenar, dorsal surface of the middle finger, and volar aspect of the forearm, were measured with RS, and different spectral patterns were found [69]. RS was also used in skin cancer diagnosis on ex vivo skin tissues, including both melanoma and NMSC [70, 71]. A more recent clinical study on in vivo skin cancer diagnosis has been conducted by our research group using an integrated real-time Raman spectroscopy system [61]. In this study, 518 skin lesions covering 8 types of common skin diseases including melanoma, BCC, SCC, actinic keratosis, atypical nevus, melanocytic nevus, blue nevus, and seborrheic keratosis were measured. Classifications using principal component with generalized discriminant analysis (PC-GDA) and partial least-squares (PLS) were carried on three paired categories, which are (1) cancerous and precancerous skin conditions (cancer plus actinic keratosis) versus benign skin lesions (noncancer); (2) malignant melanoma versus nonmelanoma pigmented skin lesions; (3) malignant melanoma versus seborrheic keratosis. Overall, with a fixed sensitivity level at  13  95%, the specificity of the classifications performed on the above-mentioned three paired categories ranges from 41% to 54%. These results on one hand demonstrated the capability of Raman spectroscopy for skin cancer diagnosis. On the other hand, it indicated the relatively low specificity of macro-Raman spectroscopy for clinical diagnosis. Further improvement is still desirable. The above studies have clearly demonstrated the broad clinical potential of RS. However, most of the in vivo Raman spectra have been collected with fiber optic probes that sample from macroscopic tissue volumes. This leads to non-specific tissue measurement and difficulties in interpreting the contributions from individual tissue layers or specific microstructures. Applying the confocal principle in microscopic imaging dramatically improves spatial resolution from millimeters to micrometers. A few groups have developed confocal Raman spectroscopic systems in the past decade. For example, water concentration in the stratum corneum has been measured with a confocal Raman system, and the results were consistent with data previously published from X-ray microanalysis of skin samples. Concentrations of the major constituents of natural moisturizing factor (NMF), lactate and urea were also determined [72]. Confocal RS was also used to monitor the delivery of compounds such as trans-retinol into human skin in vivo [73] or monitor certain penetration enhancers in human stratum corneum in vivo [74], which could significantly reduce the need for skin excision when monitoring the topical application of pharmaceutical agents. A confocal Raman microprobe was developed and utilized in vivo to determine water content changes after application of a moisturizing enhancer [75]. Significant changes were found between control and treated groups by calculating the ratio of water to protein in the confocal Raman spectra.  1.3.4 Advanced optical imaging Comparing with spectroscopy, imaging provides direct morphological and structural information of the skin. Many imaging techniques also have the merits of fast speed and the  14  similarity with histopathology results, which have great potential to be employed clinically to help skin diagnosis and evaluation. Optical coherence tomography Optical Coherence Tomography (OCT), which was developed in late 1980s and was initially used in ophthalmology, is another promising non-invasive technology for diagnosis of skin diseases. The principle of OCT is analogous to pulse-echo ultrasound imaging, but OCT uses light instead of sound to create images. Usually, low-coherent near-infrared light at wavelength around 1300 nm is used to minimize the absorption caused by protein, water, hemoglobin and lipids inside the tissue. The reflected light signals carry information about the internal microstructures within biological tissues due to their optical index differences. The echo time delay of reflected light waves is determined by an interferometer, and is converted into a two-dimensional spatial image. The intensity of the reflected light waves is translated into an intensity map. In traditional OCT systems (time-domain OCT), a moving reference mirror is usually used to calibrate the reflected light waves for image acquisition. However, the mechanics of the reference mirror is relatively slow. More recently, a second-generation technology called frequency-domain OCT (also named as Fourier-domain OCT) has emerged and replaced the time-domain OCT. Compared with time-domain OCT, frequency-domain OCT has about 3-fold better spatial resolution and 10 times faster image acquisition [76]. Frequency-domain OCT detection can be performed in two ways: spectral OCT, which uses a spectrometer with a multichannel analyzer [77–79]; or swept source OCT, which uses a fast-tunable laser source [80, 81]. OCT has been introduced in dermatology due to its high depth of penetration in 1997 and is increasingly employed in clinical skin research [82, 83]. So far, OCT has been used to study skin tumors [84], inflammatory skin diseases [85], and also to evaluate treatment effects [86]. However, the optical resolution of OCT system is relatively low (usually  15  Light Source Illumination Pinhole  Detector  Detection Pinhole  Beamsplitter  Microscope Objective Lens  Focal Plane Sample Figure 1.5: Configuration of a reflectance confocal microscope. around 10μ m ) and may not be able to differentiate different cell types and visualize subtle changes associated with early stage diseases. Confocal scanning laser microscopy Confocal scanning laser microscopy (CSLM), which is able to non-invasively image human skin in vivo in real time and with cellular resolution, has recently gained its popularity in dermatology applications. The principle of confocal imaging was first proposed in 1957 by Marvin Minsky [87]. The idea is to use a point illumination and a spatial pinhole to reject the out-of-focus light  16  in the samples, which allows imaging the structure at the focal plane only. This provides micron-scale resolution images with the capability of optical sectioning. An example configuration of a reflectance confocal microscopy is shown in Figure 1.5. The major components of a CSLM include (1) light source, which is usually a laser radiating UV, visible or NIR light; (2) scanner, which scans either the laser beam or the sample stage to be able to form a image; (3) lens or microscope objective, which is used to focus the light onto the sample; (4) detector with pinhole in front to reject the out-of-focus light and collect those light signals coming from the focal plane. There are two basic modalities of CSLM, namely fluorescence confocal microscopy (FCM) and reflectance confocal microscopy (RCM). The fluorescence confocal microscopy targets on either endogenous fluorophores inside the tissue, or the exogenous fluorophores that can be accumulated in specific structures of the tissue. The weak signals of endogenous fluorophores of biological tissues and the safety limitations on maximum laser exposure make it impractical to perform autofluorescence based FCM on human tissue in vivo. For exogenous fluorophore based FCM, the possible invasiveness and complexity of the procedure of administrating exogenous fluorophores makes it less desirable for in vivo applications. In comparison, reflectance confocal microscopy uses differences in refractive indices as its imaging contrast, which does not involve any labeling. Since RCM collects the reflectance signals from the tissue, the signal level is also higher than fluorescence mode. Therefore, RCM is suitable for in vivo applications, and enables real-time imaging of living tissue at cellular level resolution without physical tissue dissection or disruption. The first confocal scanning laser microscopy (CSLM) system was developed in 1995 and has successfully obtained cellular resolution images of human skin in vivo [88]. Subsequently, CSLM was successfully applied in imaging a variety of in vivo biological tissues. A video-rate CSLM system has been built and in vivo oral mucosa tissue has been imaged showing epithelial cells and connective tissues [89]. In vivo cervical epithelium has been imaged with a fiber optic confocal reflectance microscopy [90]. Both cell nuclei and cell  17  boundaries can be seen in the superficial and intermediate layers of the epithelium. CSLM has also been used to visualize the spleen [91], corneal [92], lung [93], and kidney [94] in in vivo murine species. Skin is the most easily accessible organ, and there have been many studies on in vivo human skin with reflectance confocal microscopy. Confocal images of the normal skin show cellular and nuclear details in the epidermis, collagen and elastic fibers in the dermis, and circulating blood cells within dermal capillaries [95]. Since melanin provides great contrast in in vivo CSLM of human skin [88], CSLM has been used to study a variety of pigmented skin lesions. Distinct features of benign and malignant melanocytic lesions have been found [96]. Other than applications in diagnosing pigmented lesions, the high resolution imaging features of in vivo RCM also facilitated diagnosis of actinic keratoses [97], allergic contact dermatitis [98], and basal cell carcinoma [99] in vivo. Because of the capability of noninvasively imaging the skin with high resolution, CSLM was also used in tumor margin mapping by comparing the confocal images to guide the Mohs Surgery [100]. Since Raman spectroscopy can provide biochemical information, and CSLM could aid in noninvasive morphological imaging of in vivo skin, a few groups have tried to combine the two techniques together to achieve high quality imaging and spectral measurements. Raman spectroscopy derived in vivo concentration profiles of water and of natural moisturizing factor for the stratum corneum along with corresponding confocal images of the skin have been shown [101]. The same study also acquired Raman spectra and confocal images of in vivo sweat ducts, sebaceous glands, and capillaries. However, in this study, the confocal Raman spectral measurements and the reflectance confocal imaging employed two different lasers with two different wavelengths, which made the registration between the confocal image and the Raman spectrum problematic. More recent studies implemented these considerations. Co-registration of the sampling locations of the imaging and spectral channels were performed [102]. However, one limitation with this study is that the confo-  18  cal Raman spectral measurement was from a single point, which may vary a lot during the acquisition period due to movement from the subject. Multiphoton microscopy There is another advanced imaging technique called multiphoton microscopy (MPM), which has comparable or better image resolution than CSLM and inherent optical sectioning capability. MPM uses NIR light as excitation to excite specific molecules and provides better penetration in highly scattering tissues such as skin. Under the excitation of a femtosecond (fs) pulsed laser source that has a high peak power, a molecule can simultaneously absorb two or more NIR photons and achieve the nonlinear excitation effect. Multiphoton excitation employs a concept first described in Maria Coeppert-Mayer’s doctoral dissertation in 1931 [103]. Since the phenomenon requires extremely high power density, it can only happen at the focal point of the excitation, which gives MPM the inherent optical sectioning capability without the need of a detection pinhole as in CSLM. This dramatically increased the number of signal photons collected by the detector, enabling tissue autofluorescence based “confocal” imaging of skin tissue in vivo, which was impractical with one-photon excitation based CSLM. The two major signal sources of MPM are two photon fluorescence (TPF) and second harmonic generation (SHG). The energy diagram showing the process of TPF, SHG and other accompanying physical processes can be found in Figure 1.6. A molecule can absorb two NIR photons within a narrow temporal window (usually in the range of 10−15 s) and is excited to a higher energy state if each of the two photons provides half the energy for the molecular transition. When the molecule returns to the ground state, the emitted photon will have a much shorter wavelength usually in the visible range. This process is called two photon fluorescence (TPF). If the molecule interacts with two photons and is excited to a virtual energy state, the wavelength of the emitted photon will be exactly half compared to the wavelength of the excitation photons. This process is called second harmonic generation (SHG). There are endogenous fluorophores  19  Real excited states  Virtual state S1  Vibrational relaxation  Intersystem crossing h n ex  h n ex Absorption of two photons via a virtual state h n ex  Interaction of two photons via a virtual state  T1 hn TPF Internal conversion  Internal conversion  Phosphorescence  hn SHG  h n ex  S0  Ground state S0  Second Harmonic Generation  Two-Photon Fluorescence  Figure 1.6: Diagram of some physical processes relevant to two photon fluorescence (TPF) and second harmonic generation (SHG). A molecule can absorb two photons and gets excited to a higher electronic state. Following vibrational relaxation, it can return to the ground state through either internal conversion or emission of two-photon fluorescence. It may also transit to the first triplet state T1 through intersystem crossing, and return to the ground state through either internal conversion or emission of phosphorescence. A molecule can also interact with two photons and gets excited to a virtual energy state. When it returns back to the ground state, the emitted photon will have exactly half wavelength compared to the excitation photons. This process is called second harmonic generation. inside biological tissues which can generate TPF signals such as keratin, melanin, reduced nicotinamide adenine dinucleotide (phosphate) (NAD(P)H), flavin, porphyrin, and elastin [104, 105], while collagen are noncentrosymmetric structured molecules which can provide strong SHG signals [106, 107]. Therefore, the advantages of MPM over CSLM include: (1) there is no need to use pinholes in MPM, which simplifies the instrumentation set-up and minimizes the signal loss caused by the pinhole; (2) because of the tight volume of multiphoton excitation, the photobleaching effect and the photodamage to the non-imaged surrounding tissue are minimized compared to one-photon based confocal imaging; (3) MPM can provide specific contrast from SHG, which contains specific structural information of the involved molecules and improves its biochemical specificity; (4) by tuning the excitation wavelength, different fluorophores inside the tissue could be excited allowing us to focus at specific molecules at a time; and (5) different from RCM, which can only gener20  ate morphological information based on refractive index differences of the sample, MPM is able to provide both morphological and biochemical information of biological tissues, and may also provide functional information when monitoring dynamic processes. Therefore, it has unique advantages for clinical applications. The applications of MPM in biological fields are increasing. For example, MPM has been used for monitoring in vivo neuron activities in mammalian brain tissue without damaging the skull [108]. In vivo immune responses can also be studied using MPM by imaging lymphocytes in intact tissues [109]. There has also been study on quantitative determination of oxygen concentration of in vivo rat kidney using two-photon excitation phosphorescence lifetime measurements [110]. MPM has also been employed to study the cellular metabolic oxidation and reduction states of rabbit cornea in situ, and high quality autofluorescence images of basal epithelium with details of cellular structures have been shown [111]. Since the first in vivo human skin images using multiphoton microscopy (MPM) were reported [112], MPM applications in dermatology have expanded dramatically. High resolution multiphoton fluorescence images and SHG images of in vivo human skin were acquired, and both cellular structures and connective tissue structures can be clearly seen [113]. MPM can also be used for studying drug delivery or accumulation of cosmetic components [114]. A combination of TPF and SHG imaging can be applied to study extracellular matrix proteins including collagen and elastic fibers, and the degree of photoaging can be quantified with SHG to autofluorecence aging index of dermis (SAAID) [115]. The exploration of MPM as an in vivo clinical skin diagnostic tool is also expanding, with commercial in vivo MPM instruments presently being available (DermaInspect and MPTflex, JenLab GmbH, Jena, Germany) in addition to several other reported systems [112, 116– 118]. Cancer cells tend to have a large cell body with low intracellular autofluorescence signals under MPM, and the decreased autofluorescence may be due to the replacement of normal cytoplasm by intracellular mucin [119]. MPM was also used to explore the feasibility of differentiating nonmelanoma skin cancer from normal skin, and significantly different  21  morphologic features were found between BCC lesions and perilesional skin [120]. Ex vivo BCC has been visualized using MPM. SHG signals were found to be decreased in the cancer stroma compared to normal dermis, while the autofluorescence signals increased [121]. Squamous cell carcinoma in situ and superficial basal cell carcinoma were also imaged using MPM, and significant features were found between lesions and perilesional skin [120]. A study on using MPM for diagnosing malignant melanoma has shown that with seven unique features derived from the autofluorescence images, a sensitivity of 95% and a specificity of 97% can be achieved [122]. Other than applications in diagnosis, realtime monitoring for certain biological process is another area that MPM can be involved. Concentration of a fluorescent drug inside the skin was monitored for more than 5 hours using MPM, which demonstrated the potential of MPM for studying drug penetration and delivery at microscopic level [123]. However, one limitation with the current MPM systems is the slow scanning rate, which may cause image blurring due to subject movement and increases the time for imaging multiple lesions and for performing volumetric imaging. Therefore, a system that could provide fast, video-rate imaging speed will help with these problems. Moreover, the hypothesis of this thesis is that if MPM system can be integrated with a reflectance confocal imaging module and a confocal Raman spectroscopy module, it should provide complementary and rich information for better skin characterization and evaluation.  1.4 Objective and outline of this dissertation The objective is to design and develop a multimodality optical instrument, which could provide both accurate biochemical and morphological information of human skin in vivo and aid in noninvasive skin diagnosis and evaluation. Since Raman spectroscopy (RS) is a vibrational spectroscopic technique capable of providing details on the biochemical composition, molecular structure, and molecular interactions in cells and tissues, it was selected as one of the modalities of the designed instrument. To obtain morphological information of  22  the skin, reflectance confocal microscopy (RCM), which can generate micron-level image resolution with the capability of optical sectioning, was employed as the second modality. Multiphoton microscopy (MPM) is another advanced imaging method, which uses nonlinear near-infrared excitation to excite specific molecules inside the tissue and provides subcellular level image resolution and inherent optical sectioning capability. Moreover, because MPM can provide biochemistry based morphological information simultaneously from the two-photon fluorescence (TPF) and second harmonic generation (SHG) images, it was included as the third modality of the designed system. The hypothesis is that a multimodality instrument that is able to acquire confocal Raman spectra, reflectance confocal images, and multiphoton microscopic images of in vivo human skin can be developed as a tool for noninvasive skin diagnosis and evaluation. The thesis project was completed in six steps, which are described in the six research chapters from Chapter 2 to Chapter 7: The first step of the whole study was to design and develop a confocal Raman spectroscopy system, which is the essential basis for obtaining biochemical information of the skin. The system development, evaluation, and a mouse skin study are described in Chapter 2. The second step, which is a significant move towards the final goals, was integrating a reflectance confocal imaging modality into the developed confocal Raman spectroscopy system to achieve acquisition of both biochemical and morphological information of the skin. The system development, tests on ex vivo and in vivo skin are presented in Chapter 3. Chapter 4 describes the third step of this study, which is to design and construct a video-rate multiphoton microscopy system to acquire both morphological and structural information of the skin. Both ex vivo and in vivo human skin were imaged with the developed multiphoton microscopy system, and three-dimensional volumetric measurements were also performed. The fourth step was combining the reflectance confocal imaging module with the de-  23  veloped multiphoton microscopy system to provide more information for skin diagnosis (Chapter 5). Imaging with the developed co-registered femtosecond reflectance confocal channel and the multiphoton microscopy channel was performed on human skin. The next step was integrating the confocal Raman spectroscopy module into the combined reflectance confocal and multiphoton microscopy system to provide biochemical, morphological and functional information of skin within one instrumentation (Chapter 6). A study to acquire confocal Raman spectra, continuous-wave reflectance confocal images, femtosecond reflectance confocal images, and the combined two-photon fluorescence and second harmonic generation images in vivo from volunteers was performed (Chapter 6). The last step was demonstrating an example application of the developed system, which was two-photon absorption based precisely targeted photothermolysis on the skin monitored by the developed reflectance confocal and combined two photon fluorescence and second harmonic generation imaging modalities. The results obtained demonstrated unique capabilities of the multimodality system and supported the overall hypothesis. Chapter 8 summarizes the conclusions of this dissertation and discusses the future directions. To our knowledge, this is the first time to report a multimodality imaging and spectroscopy system which could noninvasively obtain co-registered reflectance confocal microscopy (RCM), two-photon fluorescence (TPF) and second harmonic generation (SHG) images simultaneously at video-rate, as well as perform image-guided region-ofinterest Raman spectral measurements of human skin in vivo. Future directions include the further exploration of the capability of the developed system in skin disease diagnosis and lesion margin assessment by measuring more skin patients; to further optimize the irradiation protocol and perform more tests on human skin samples in order to better adapt the two-photon absorption based light therapy into clinic; improve the probe design to be able to access more body sites and to further reduce the effects of involuntary body movement from the subjects.  24  Chapter 2 In Vivo Confocal Raman Spectroscopy A version of Sections 2.2 and 2.4 in this chapter have been published as: Wang, H., Huang, N., Zhao, J., Lui, H., Korbelik, M. and Zeng, H. (2011), Depth-resolved in vivo micro-Raman spectroscopy of a murine skin tumor model reveals cancer-specific spectral biomarkers. Journal of Raman Spectroscopy, 42: 160-166. A version of Sections 2.5 in this chapter has been published as: Wang, H., Zhao, J., Lee, A.M.D., Lui, H., and Zeng, H. (2012), Improving skin Raman spectral quality by fluorescence photobleaching. Photodiagnosis and Photodynamic Therapy, 9: 299-302. Edits have been made to better integrate these publications into the flow of the thesis.  2.1 Introduction Raman spectroscopy (RS) has been used in a variety of applications in biology [42] because it is relatively simple, reproducible, nondestructive to the tissue, and has minimal sample preparation requirements. RS also provides specific and data-rich information of major components of biological tissues, such as blood, lipids, proteins, and nucleic acids [39, 43]. However, most of the studies so far have been focused on obtaining biochemical information from a large volume, which leads to non-specific tissue measurement and difficulties in interpreting the acquired Raman spectra. Confocal Raman spectroscopy offers  25  cellular level resolution, and allows non-invasive acquisition of depth-resolved biochemical information of in vivo tissues. There have been some studies on using confocal Raman spectroscopic system to measure the skin. For example, concentrations of the major constituents of natural moisturizing factor (NMF), lactate and urea have been determined [72]. Water content changes after application of a moisturizing enhancer have also been monitored with a confocal Raman spectroscopic system [75]. However, some biological tissues such as skin may have strong fluorescence background when performing Raman spectroscopic measurements, making the process of extracting the Raman spectrum difficult. Therefore, a method of decreasing the autofluorescence is preferred. Some previous studies have shown that tissue fluorescence emitted by almost all fluorophores fades as a result of light exposure [124–126]. This process is called photobleaching. Developing a confocal Raman spectroscopy system is considered as the first step to achieve the goal of building the multimodality instrument, which integrates a confocal Raman spectroscopy module, a reflectance confocal imaging module, and a multiphoton microscopy system to realize noninvasive skin characterization and evaluation. Therefore, this chapter focuses on designing and developing a confocal Raman spectroscopy system, characterizing the system performance, and validating the system on an in vivo cutaneous murine tumor model. Another focus of this chapter is to discuss fluorescence photobleaching and its potential function for improving Raman spectral quality. The hypothesis of this chapter is that the developed confocal Raman spectroscopy system should be able to acquire depth-resolved biochemical information from in vivo mouse skin and differentiate normal mouse skin from tumor-bearing mouse skin without cutting the skin. Another hypothesis of this chapter is that fluorescence photobleaching could be used to improve Raman spectral quality.  26  2.2 System design and development A confocal Raman spectroscopy (RS) system specially designed for in vivo depth-resolved analysis of the skin was built. The system diagram is shown in Figure 2.1. The excitation source is a single-mode, stabilized diode laser (785 nm, 100 mW , Model #: I0785SU0100BTK, Innovative Photonic Solutions, Monmouth Junction, NJ, USA), which has a central wavelength of 785.71 nm and an output power of 100 mW . The laser beam first passes through a spatial filter system which consists of two lenses (focal length of 11 mm and 50 mm, respectively) and a pinhole (size of 30 μ m). The function of the spatial filter system is to first purify the laser mode and also to expand the laser beam to fully fill the back aperture of the microscopy objective. A bandpass filter is used to purify the laser wavelength. The laser beam then travels through a dichroic beamsplitter and is focused onto the skin by a water-immersion objective (Model #: LU MPL40×W /IR, numerical aperture (N.A.) 0.8, working distance (W.D.) 3.3 mm, Olympus, Markham, Ontario, Canada). The back scattered raw spectral signals, which was composed of both Raman signals and the tissue autofluorescence background are collected by the same microscope objective and directed by the dichroic beamsplitter into a low-OH (low hydroxyl content) single optic fiber of 100 μ m core diameter with a high near infrared (NIR) transmission. The long pass filter on the collection path is for blocking the reflected laser light and the lens is for focusing the Raman signals into the collection fiber. A spectrometer system which collects the Raman signals from the optic fiber, was equipped with an NIR-optimized back-illumination deep-depletion charge coupled device (CCD) array (Spec − 10 : 100BR/LN, Princeton Instruments, Trenton, NJ, USA) and a transmissive imaging spectrograph (HoloSpec − f /2.2 − NIR, Kaiser, Ann Arbor, MI, USA) with a volume phase technology holographic grating (HSG − 785 − LF , Kaiser). The CCD had a 16 − bit dynamic range and was cooled with liquid nitrogen to −120 ◦C. The spectral resolution of the system was 8 cm −1 . A specially designed probe which reduces the effects of involuntary body movements from the subjects was used in all the mouse experiment. 27  Figure 2.1: Schematic configuration of the confocal Raman spectrometer system for depth-resolved skin Raman measurements. The light intensity after the objective and incident on the tissue surface was 27 mW , and the irradiance on skin was well below the American National Standards Institute maximum permissible exposure level [127].  2.3 Resolution measurement and system calibration The resolution of an optical microscopic system is usually defined as the shortest distance between two points on a specimen that can still be distinguished by the observer or camera system as separate entities. Therefore, the resolution of a system is a crucial parameter, which characterizes the system performance. In order to learn the system performance of the developed system, both lateral and axial resolution of the system were measured. Since silicon has one strong and sharp Raman peak at 520 cm−1 , and the excitation light at 785 nm cannot penetrate the silicon, a piece of silicon wafer which has a sharp edge was manually scanned by the focused laser beam with a stepsize of 1 μ m. The Raman signals of the silicon wafer was recorded with the spectrometer. During the scan, the strongest Raman peak of the silicon, which is located at 520 cm −1 , was maximized for each measurement 28  Figure 2.2: Integrated Raman intensity versus lateral displacement on a silicon wafer. point by axially scanning the silicon wafer using a micrometer. Integrated Raman intensity under the peak at 520 cm−1 was used to plot against the lateral displacement along the silicon wafer. As shown in Figure 2.2, 0 μ m represents the first measurement point which was still on the silicon wafer and around 40 μ m away from the edge. Moving to the edge of the wafer, the integrated Raman intensity under the peak is roughly flat. When the measurement point was at the edge, which is shown as 40 μ m on Figure 2.2, the integrated Raman intensity dropped suddenly and eventually went down to 0. The measurement was continued till around 30 μ m away from the edge of the silicon wafer. The lateral resolution was then determined by the 10% to 90% edge response of a Boltzmann fit [128], which was around 2.2 μ m. To determine the axial resolution of the system, Raman measurement along the z-axis of a piece of silicon wafer was performed with a stepsize of 2 μ m. Integrated Raman intensity under the peak at 520 cm−1 was used to plot against the axial displacement [129] from 90 μ m below the silicon plane to 50 μ m above the plane. The result is shown in Figure 2.3. As expected, the integrated intensity started from 0 when the focal plane is 29  Figure 2.3: Integrated Raman intensity versus axial displacement along a silicon wafer. below the silicon plane and increased sharply when the focal plane is close to the silicon plane. The integrated intensity then decreased to 0 when the focal plane is above the silicon plane. The full width at half maximum (FWHM) of a Gauss fit was used to determine the axial resolution of the system, which is around 7.3 μ m. Wavelength calibration of the system was performed using a Mercury Argon lamp (HG − 1 Mercury Argon Calibration Light Source, Ocean Optics, FL, USA) and a Neon lamp (NE −1 Neon Calibration Light Source, Ocean Optics). Ten peaks including the laser line were selected and a 5th order polynomial fitting was used to complete the calibration curve. Because the CCD may have different response rates for difference wavelengths, an intensity calibration procedure using a standard tungsten lamp (RS-10A-1, Gamma Scientific, CA, USA) was also performed to compensate for that.  30  2.4 Measurements on in vivo mouse skin Mouse skin was selected to validate the performance of the developed system. Confocal Raman spectra were taken from the tumor-bearing mice in vivo in order to assess (1) the Raman spectral differences between different skin layers and (2) the spectral changes for both the epidermis and the dermis between normal peritumoral skin and skin immediately overlying subcutaneous tumors. All animal experiments were performed according to a protocol approved by the University of British Columbia (UBC) Committee on Animal Care. In total, 24 mice were used in this study. The squamous cell carcinoma (SCCVII) tumors were generated by subcutaneous injection of 3.6×10 6 cells in 50 ml phosphate buffered saline into the back of female C3H/HeN mice. Raman spectral measurements were performed when the tumor volume reached 90-120 mm3 (around 10 days after tumor inoculation). The dimensions of each tumor were measured by calipers every other day and their volumes were calculated with Equation 2.1, assuming each tumor is an elliptical sphere.  Volume = (π /6) × (tumor length) × (tumor width) × (tumor height)  (2.1)  All mice were shaved and anesthetized before measurement. Axial scanning from the skin surface to deeper layers was performed both at the tumor site and at a normalappearing skin site (approximately 3-4 cm away from the tumor site) within the same anatomic region. After each experiment, the skin that was measured was excised, processed for histologic examination, and the skin sections stained with hematoxylin and eosin (H&E). In total, 264 spectra from normal sites and 230 spectra from tumor sites at depths ranging from 10 μ m to 140 μ m below the skin surface were acquired.  31  Figure 2.4: Fluorescence background removal of an example Raman spectrum from mouse skin.  2.4.1 Data processing and statistical analysis Fluorescence background removal The measured Raman spectra are superimposed on top of a fluorescence background, which varies with each measurement. There are several methods for rejecting NIR autofluorescence background signals. The most commonly used method in biomedical Raman measurement is polynomial curve fitting. For biochemical components analysis, it was found that the modified polynomial curve fitting fluorescence removal method is a better choice, as shown in Figure 2.4 (details can be found in [130]). It is an iterative polynomial fitting method, which combines a peak removal procedure and takes into account noise. Figure 2.4 shows the interface of the software developed by our group, Vancouver Raman Algorithm (http://www.bccrc.ca/dept/ic/cancer-imaging/haishan-zeng-phd). Nominally pure Raman spectra can be analyzed after subtracting the fluorescence background.  32  Normalization All the background-subtracted Raman spectra were normalized to the area under curve (AUC) of each spectrum to account for the influence of inter- and/or intra-subject spectral variability on the multivariate data analysis. For convenience in displaying the spectra, the normalized intensities were multiplied by the number of data points (888) in each spectrum. Multivariate data analysis Principal component analysis (PCA) [131] was used to analyze the Raman spectra. The entire spectral range (500-1800 cm−1 ) was used for PCA by representing each spectrum as a set of top four PCA variables for epidermal spectra and top three PCA variables for dermal spectra which represented 70% of the total variance of the original data. PCA was performed on the standardized spectral data matrix to generate principal components (PCs) comprising a reduced number of orthogonal variables that accounted for most of the total variance in original spectra. Each PC was related to the original spectrum by a variable called the PC score, which represented the weight of that particular component. For dermal spectra analysis, the first three PC scores that have the largest eigenvalues were then applied to a linear discriminant analysis (LDA) model for tissue classification [132]. Similarly, for epidermal spectra analysis, the first four PC scores were used. LDA determined the discriminant function line that maximized the variance in the data between groups while minimizing the variance between members of the same group. The performance of the diagnostic algorithms rendered by the LDA models for correctly predicting the tissue status (i.e. tumor vs. normal) underlying each spectrum was estimated in an unbiased manner using the leave-one-out cross-validation method on all spectra. In this method, one spectrum was removed from the dataset and the entire algorithm, including PCA and LDA, was redeveloped and optimized using the remaining spectral set. The optimized algorithm was then used to classify the withheld spectrum and this process was repeated until each spectrum was individually classified. To compare the performance of  33  the PCA-LDA model for tissue classification using the spectroscopic dataset, receiver operating characteristic (ROC) curves were generated by successively changing the thresholds to determine correct and incorrect classification for all samples. All multivariate statistical analyses were performed using MatLab software (Version 7.6, MatLab Software, The MathWorks Inc., Natick, MA, USA) with the Statistical Pattern Recognition Toolbox (Vojtech Franc and Vaclav Hlavac, Czech Technical University Prague, Faculty of Electrical Engineering, Center for Machine Perception, Czech Republic).  2.4.2 Classification of normal mouse skin and malignant mouse skin With a 15 s integration time under 27 mW of excitation light exposure to the skin surface, Raman spectra with good signal-to-noise ratio (SNR) were obtained. Histologically, there were many more cells in the tumor compared with normal skin as shown in Figure 2.5. A total of 494 Raman spectra were taken. The mean normalized Raman spectra of normal skin from 24 mice (Figure 2.6(a)) varied significantly according to the depth. For example, strong peaks at 1061, 1128, and 1296 cm −1 attributed to ceramide [69] occurred in the spectra of the epidermis, while peaks at 855 and 937 cm −1 coming from collagen featured prominently in the dermal spectra. Obvious spectral differences between both the epidermis and the dermis were also observed between normal and tumorbearing mouse skin (Figure 2.6(b)). Figure 2.7 shows these spectral variations more clearly via a comparison of the mean normalized spectra between normal and tumor site skins. In the dermis (Figure 2.7(b)), the mean spectrum of the tumor has higher intensities for the peaks at 724 and 1093 cm−1 and the band 1325-1330 cm−1 , which arise from nucleic acids [133–135], indicating a higher density of cells or nuclei in the tumor, or increased DNA in that the individual tumor cells may have more than the usual amount of DNA in each cell. Differences in phenylalanine were also found between normal and tumor sites (Figure 2.7(a) and Figure 2.7(b)). A significant peak at 899 cm−1 was found in the tumor group but not in the normal group at the epidermal level (Figure 2.7(a)), which is also  34  (a)  (b )  Figure 2.5: Photomicrographs of mouse skin from (a) normal skin and (b) tumor (H&E). shown in Figure 2.10. Assignments of the major Raman peaks are summarized in Table 2.1. Four sets including 48 normal spectra (10 and 20 μ m depth), 48 tumor spectra (10 and 20 μ m depths), 48 normal spectra (30 and 40 μ m depths), and 48 tumor spectra (30 and 40 μ m depths) were used in the PCA. For the epidermis (10 and 20 μ m) pair, four PCs retaining 70% of the variance of the original data were kept for discriminant analysis to differentiate the tumor from normal cell. Figure 2.8 shows these four PC functions. For the dermis (30 and 40 μ m) pair, three PCs accounted for 70% of the variance and were used for analysis. Leave-one-out cross-validation procedures were used in order to prevent overtraining. The three PCs for the dermis are plotted in Figure 2.9, which shows that the 35  (a)  (b)  Figure 2.6: Mean normalized spectra for normal and tumor site skin at different depths from 24 mice: (a) mean normalized spectra for normal skin at depths of 10, 20, 40, 70 μ m (average of 19 spectra) and 100 μ m (average of 10 spectra); (b) mean normalized spectra for tumor site skin at depths of 10, 20, 40, 70 μ m (average of 17 spectra) and 100 μ m (average of 2 spectra). Each spectrum is an average of 24 spectra obtained at each depth except for the 70 and 100 μ m depths where the number of spectra used for averaging is indicated in parentheses above. PCs picked up the information coming from collagen (855 and 937 cm −1 ), phenylalanine (1001 cm−1 ), lipids (1061, 1128, and 1296 cm −1 ), and nucleic acids (1325-1330 cm−1 ). This is in good correlation with the major differences observable in the spectra between 36  (a)  (b )  Figure 2.7: Comparison of mean normalized spectra between normal skin and tumor site skin at the epidermis and dermis: (a) mean normalized skin spectra at the epidermal layer (10 and 20 μ m depths) from 24 mice; (b)mean normalized skin spectra at dermal layer (30 and 40 μ m depths) from 24 mice. Each spectrum is an average of 24 spectra obtained at each depth from the 24 mice. Dotted lines represent the most significant differences on the mean spectra. normal and tumor groups in the dermis. In the epidermis, the PCs also picked up the 899 cm−1 signal, which is the most significant difference between normal and tumor-bearing skin (Figure 2.10). Figure 2.11 shows the scatter plot of the three PC scores (PC1, PC2, and PC3) for all the dermal spectra, demonstrating that the two groups (normal skin vs tumor)  37  Table 2.1: Raman peak assignment for ex vivo mouse skin Peak position (cm−1 ) Assignment Reference 538 Cholesterol ester [136] 572 Tryptophan/cytosine, guanine [137] 724 Nucleic acids [133] 759 Tryptophan [137] 817 C-C stretching (collagen assignment) [44] 855 Collagen [138] 937 Collagen type I [138] 1001 Phenylalanine [134] 1061 Lipids [139] 1093 Symmetric PO2− stretching vibration of the DNA [133] 1128 ν (C-C) skeletal of acyl backbone in lipid (transformation) [140] 1176 C-H bending tyrosine (proteins) [140] 1206 Hydroxyproline, tyrosine (collagen assignment) [137] 1247 Amide III (collagen assignment) [44] 1267 C-H, amide III [134] 1296 CH2 deformation (lipid) [139] 1325-1330 Nucleic acids [135] 1444 Protein and lipids [141] [142] 1650 Amide I (protein) [134] 1742 C=O, lipids [137]  Figure 2.8: Four PCs from PCA analysis of spectra at the epidermal level.  38  Figure 2.9: Three PCs from PCA analysis of spectra at the dermal levels.  Figure 2.10: Comparison of mean normalized spectra for normal and tumor site skin at the epidermal level with standard deviation. can be very well separated. An optimal diagnostic sensitivity of 95.8% and specificity of 93.8% were observed. The area under the ROC curve was 0.96 (Figure 2.12). For the epidermal spectra, an optimal sensitivity of 89.6%, specificity of 89.6%, and an AUC of 0.88 were obtained. 39  Figure 2.11: Scatter plot of three PC scores (PC1, PC2, and PC3) at the dermal levels. The red triangles represent tumor group and the black dots represent normal group.  Figure 2.12: ROC curve generated from the PCA-LDA analysis of the dermal spectra using the three PCs (AUC = 0.96). Visual inspection of the spectra was also performed to test the results. Two criteria were used in the classification. The first criterion was the peak at 899 cm−1 for spectra at 40  the epidermis level. Two normal spectra showing this peak and two tumor spectra without the peak were found, which yields a sensitivity of 95.8% and a specificity of 95.8%. The second criterion was the ratio (R) of the integrated intensity from 1240 to 1269 cm −1 to the integrated intensity from 1269 to 1340 cm −1 for the spectra at the dermis level representing information from the nucleic acids. There were nine normal spectra showing a ratio smaller than 1, which means that higher concentrations of nucleic acids were present, and two tumor cases showing a ratio larger than 1, which means that lower concentrations of nucleic acids were present. A sensitivity of 95.8% and a specificity of 81.3% were observed.  2.4.3 Discussion and conclusion The axial scanning results revealed significantly different spectral patterns as a function of depth for both normal skin and tumor. For example, in Figure 2.6(a) the first two spectra (10 and 20 μ m) show very high intensities at 1060, 1124, and 1296 cm −1 , which were assigned to ceramide presumably within the stratum corneum according to a previous work.[26] The relatively high intensity at the 1325-1330 cm −1 band was derived from epidermal cells. The spectral pattern changed for depths greater than 30 μ m, i.e. within the dermis of the skin. The characteristics of dermal spectra were the two collagen peaks at 855 and 937 cm−1 and a lower intensity at the 1325-1330 cm −1 nucleic acid band, which is due to the less cellular nature of the dermis. In Figure 2.6(b), a peak at 899 cm−1 , which was consistently present at the epidermis level in the tumor group but not in the normal group was observed. Using this signal as a tumor marker, a diagnostic sensitivity of 95.8% and specificity of 95.8% was achieved. From a search of the Raman literature on tissue, one paper from Stone’s group that reported a small shoulder band in the Raman spectrum near 899 cm −1 from excised human laryngeal tissue was found [53]. They assigned this shoulder band to C-C stretching mode of proteins (α − helix conformation) without giving detailed reasoning. The reason why better signals of this peak were seen in the thesis study may be because the measurements of this thesis  41  study was performed on in vivo mouse skin, while the measurements in the literature was conducted on ex vivo tissue samples. It was found that this peak is very close to a major peak of proline at 892 cm−1 and a shoulder band of oleic acid at 897 cm−1 according to the previous RS study of a variety of biomolecules (Zhao J, Lui H, McLean DI, and Zeng H, Personal Communication). This leads to two possibilities for the peak assignment. First, it is possible that the content of proline is increased in cancerous tissue compared to normal tissue. De Heller-Milev et al. [143] have reported the differential expression of small proline-rich proteins (SPRRs) in neoplastic, inflammatory, and normal skin. SPRR is a family of the key proteins involved in the formation of the cornified cell envelope during the late stages of epidermal differentiation, which in turn is essential for epidermal barrier function and protecting the body against environmental attack and water loss. SPRRs were expressed in terminally differentiating human keratinocytes and were first identified as products of ultraviolet-C induced mutations in human keratinocyte cultures [144–146]. In references [145–147], they found increased suprabasal SPRR1 and SPRR2, but no SPRR3 expression in inflammatory dermatoses with orthokeratotic and parakeratotic squamous differentiation. By contrast, differentiating epidermal tumors such as squamous cell carcinoma (SCC) expressed SPRR3. There is also evidence showing that in normal epidermis, positive staining for SPRR was observed in keratinocytes in the granular layer and the uppermost or two spinous cell layers, whereas in SCC, staining was observed in keratotic cells around horny pearls, which indicates a difference in localization of SPRR between the two conditions [147]. Thus, it is perhaps not surprising that more obvious Raman signals of SPRR were observed in the SCC lesion than in normal skin. Another possibility is that the signal may have arisen from a higher fatty acid concentration in the cancerous tissue. Fatty acids play a significant physiological role in the stratum corneum (SC) and the free or esterified fatty acids in the human SC have chain lengths varying from 14 to 24 carbon atoms, predominantly C16 and C18 chains [148]. Louw et al. investigated the role of essential fatty acids and their metabolites during cervical carcino-  42  genesis and observed higher levels of oleic acid in infiltrating lesions compared to normal tissue. Oleic acid was significantly higher in cervical carcinomas compared with normal tissue (P < 0.01) and it was significantly lower in normal tissue compared to cervical intraepithelial lesion [149]. A threefold elevated prostatic palmitoleic acid was also found in patients with prostate cancer compared to those with benign hyperplasia in Mamalakis’s study, indicating a possible role of the fatty acid in neoplastic processes [150]. There are also studies showing significant decreases in the stearic to oleic acid ratio in the erythrocyte membrane (saturation index) in patients with advanced carcinoma of gallbladder [151], colorectal carcinoma [152], bronchogenic carcinoma, and lymphoma [153] compared to healthy patients. Pandey et al. [151] obtained a sensitivity of 100% and specificity of 77.7% using both the erythrocyte saturation index and age as a means of differentiating between controls and primary gallbladder carcinoma. This decreased saturation index may empirically indicate an increasing concentration of oleic acid, which may be useful in the diagnosis and postoperative monitoring of patients with cancer. So far, firm conclusion could not be drawn, and the exact assignment of this peak at 899 cm −1 deserves further investigation. When comparing the differences between tumor-bearing and normal skin, tumor spectra showed more obvious variations from normal spectra at 30 and 40 μ m depths than at 10 and 20 μ m depths. This was probably due to the fact that there were fewer tumor cells in the upper layers superficial to the bulk of the tumor. As shown in Figure 2.7(b), a much higher intensity at 1001 cm −1 , which represents phenylalanine was observed. This is in accordance with previous results from other groups who found that the tumor tissue contained higher concentrations of this type of amino acid [141, 154, 155]. Another significant difference between the cancerous tissue and normal skin at the dermal level was the concentration of nucleic acids represented by the three peaks at 724, 1093, and 1325-1330 cm−1 . This was due to the much higher concentration of cells in the tumor than in the normal skin, which could also be clearly seen in the H&E stain results (Figure 2.5).  43  The in vivo macro-Raman data from the human clinical study using a fiber optic probe performed by our group was reviewed. It was noted that for SCC, actinic keratosis, and psoriasis, all of which involve hyperkeratinization and may therefore have more SPRR, 21, 26, and 24% of cases showed the peak around 899 cm−1 , respectively (unpublished observations). However, this same peak was present in only 2% of the basal cell carcinoma (BCC) spectra, which may confirm the relevance of the peak at 899 cm−1 as being related to keratinization since BCC does not tend to have excess visible scale or hyperkeratinization as compared to the other conditions. The macro-Raman system was designed to measure a relatively large volume of lesion (3 mm size) using a fiber optic probe and hence may have less strong capability for differentiating different types of lesions. In macro-Raman measurements, specific Raman bands (e.g. at 899 cm−1 ) originated from thin tissue layers, or certain microstructures might have been overwhelmed by the integrated Raman signals from the large volume of the measurement tissue, while depth-resolved confocal microRaman measurements would overcome this problem. Therefore, the micro-Raman system has the potential to generate a much higher accuracy in detecting specific spectral changes and greatly improve the clinical diagnosis of different skin diseases. In summary, the developed confocal Raman system demonstrated in vivo variations in chemical composition in the skin according to depth and the presence or absence of neoplasia. The consistent identification of specific Raman spectral changes may serve as a basis for improving noninvasive optical skin cancer detection, and also points to the utility of confocal RS as a clinical tool for in vivo human skin biochemical analysis and disease diagnosis.  2.5 Raman spectral quality improvement One practical problem with both the conventional RS and the confocal RS is that the fluorescence background from the tissue is sometimes overwhelming and makes the process of extracting the Raman spectrum extremely difficult. In regular clinical RS measurements,  44  the strong autofluorescence from certain body sites, such as forehead and nose, may even saturate the spectrometer, which makes it impossible to obtain valid Raman spectra even with a short, 1 s detector integration time. Shortening the integration time too much will result in poor signal-to-noise ratio (SNR) of the extracted Raman spectra after fluorescence background removal due to the inherent weakness of the tissue Raman signals. It is known that fluorescence emitted by almost all fluorophores fades as a result of light exposure in a process called photobleaching or photodegradation [124–126]. Photobleaching has been exploited since the mid-1970s [156, 157] for producing valuable information about biological system dynamics. In general, it is believed that photobleaching process involves a photochemical modification of the fluorophore resulting in either lower fluorescence quantum yields or different excitation and emission wavelengths [158, 159]. Photobleaching is usually considered as an irreversible phenomenon and the fluorescence recovers only because fluorescent molecules diffuse into the bleached sample volume [159, 160]. There are many factors, such as the molecular environment and the intensity of excitation light, which may affect the mechanism, and thus the reaction order and the rates of photobleaching [161]. The kinetics of skin autofluorescence and photobleaching during Raman spectral measurements has not been fully characterized. Therefore, the autofluorescence decay process for both ex vivo mouse skin and in vivo volunteers was studied using the home-made macro-RS system developed by our group and the confocal RS system that was discussed in the previous sections of this chapter. The goal was to develop a novel method to reduce autofluorescence of skin and to improve quality of Raman spectra in clinical measurements.  2.5.1 Materials and methods Confocal Raman spectroscopy system and ex vivo mouse skin measurement procedure For the ex vivo mouse skin measurements, the same confocal Raman spectroscopy system that was discussed in previous sections of this chapter with some modifications was uti-  45  lized. Instead of using an 100 μ m optic fiber, a 50 μ m one was used, which improves the spectral resolution from 8 cm−1 to 4 cm−1 . To create different power density at mouse skin, the beam was scanned by a resonance scanner operating at 8 kHz for the fast axis, and galvanometric scanning for the slow axis (CRS8 and VM500+, Cambridge Technology, Lexington, MA). Confocal Raman spectral measurements on three ex vivo mouse skins were performed. Before each experiment, the hair on the back skin of the mouse was removed with a razor and the mouse was euthanized with carbon dioxide (CO 2 ). The skin was excised and was attached onto a microscope glass slide. For the scanning mode measurements, different scanning areas, including 30μ m×30μ m, 70μ m×70μ m, 150μ m×150μ m, 300μ m ×300μ m, and 500μ m ×500μ m were utilized, and the corresponding power density is 3 × 103 W /cm2 , 550 W /cm2 , 120 W /cm2 , 30 W /cm2 , 10 W /cm2 , respectively. In order to record the continuous fluorescence change during the whole measurement, the exposure time of each spectrum was set to be 0.3 s. Clinical Raman spectroscopy system and in vivo volunteer skin measurement procedure A home-made real-time rapid Raman spectroscopy system was employed for the in vivo macro-Raman measurements. Detailed information about this system can be found elsewhere [162]. In total, three body sites including inner forearm, dorsal forearm, and forehead on three volunteers (2 Asian males aged 40-50 and 1 Asian female aged 26) were measured. The forearm was measured by placing it on top of a lab jack, the distance between the forearm skin and the measuring probe was adjusted and fixed. An opaque block was placed in between the probe and skin to avoid light exposure before the spectral measurement, and it was removed when the measurement starts. For the forehead measurement, the probe was held in place by an assistant and all the other procedures were similar as for the forearm measurement. Before each measurement, the skin was cleaned with alcohol pad to remove make-up or other residuals on the skin surface. A plastic ring was utilized to mark the measuring site. In order to characterize the fluorescence decay parameters, 100 s 46  continuous spectral measurements with 1 s detector integration time on the inner forearm was recorded. The forehead was also measured continuously and with different detector integration times varying from 0.5-2 s. The measured raw spectra were normalized to the integrated laser peak intensity to removal the effect of laser power fluctuations on the spectral signal. Both the Raman spectra and the autofluorescence spectra were extracted from the raw data with 5th order polynomial fittings [130]. Informed consent was obtained from each volunteer subject.  2.5.2 Results The fluorescence photobleaching kinetics in laser scanning confocal Raman measurement of ex vivo mouse skin was first studied. The fluorescence intensity is integrated from 777 nm to 923 nm. Figure 2.13(a) shows the mouse skin fluorescence decay curves at 10 μ m depth. As expected, the fluorescence decay slows down with increasing scanning area because the average illumination power density decreases. It was found that, there were more than 50% fluorescence loss at 60 s for the single point measurement, around 40% fluorescence loss at 60 s for 30μ m × 30μ m measurement, around 35% fluorescence loss at 60 s for 150μ m × 150μ mm measurement, less than 25% fluorescence loss at 60 s for 300μ m × 300μ m measurement, and fluorescence loss for 500μ m × 500μ m scanning area at 60 s is even less than 10%. The fluorescence decay kinetics at different depths inside the skin are also studied. Figure 2.13(b) and Figure 2.13(c) show the fluorescence decay curves at 30 μ m and 60 μ m in depth, respectively. It was found that the single point measurement still photobleaches the most when compared with scanning measurements. Fluorescence decay curves follow the same trend as the 10 μ m depth measurement. But for the same scanning area, the fluorescence loss at 30 μ m depth is smaller than at 10 μ m depth and the fluorescence loss at 60 μ m depth is even smaller. To better illustrate the depth effect, the autofluorescence decay kinetics at three different depths for the single point measurement is shown in Figure 2.13(d). More than 50% fluorescence loss at 10 μ m depth was found  47  (a)  (b )  (c)  (d)  Figure 2.13: Ex vivo mouse skin fluorescence decay curves with different scanning areas and at different depths. Scan areas include: nonscan, 30μ m × 30μ m, 150μ m × 150μ m, 300μ m × 300μ m, 500μ m × 500μ m. (a) different scanning areas at 10 μ m depth from the surface of the skin; (b) different scanning areas at 30 μ m depth from the surface of the skin; (c) different scanning areas at 60 μ m depth from the surface of the skin; (d) different depths for nonscan mode. The power densities for different scanning areas are: 2.15 × 105 W /cm2 (nonscan mode), 3 × 103 W /cm2 (30μ m × 30μ m), 550 W /cm2 (70μ m × 70μ m), 120 W /cm2 (150μ m × 150μ m), 30 W /cm2 (300μ m × 300μ m), and 10 W /cm2 (500μ m × 500μ m), respectively. while less than 30% loss at 60 μ m depth after 60 s was found. In order to study in vivo human skin fluorescence photobleaching characteristics, the macro-Raman experiments was performed using the clinical Raman system. It was found that the raw signals, including contributions from both the fluorescence and the Raman signals, decrease with time as shown in Figure 2.14(a). Interestingly, Figure 2.14(b) illustrates that the fluorescence decreases monotonically with time, while the Raman signal stays al48  most constant (Figure 2.14(c)). It was also find that with an incident power density of 0.29 W /cm2 (below the ANSI standard MPE of 0.3 W /cm2 for 785 nm laser) at the skin surface, fluorescence decreased approximately 25% at 200 s exposure time (Figure 2.14(d)). The decay curve can be well fitted to a double exponential curve with a fast decay component (time constant, 2.3 s) and a slow decay component (time constant, 77.4 s) as shown in Equation 2.2:  I(t) = 0.12exp(−t/2.3) + 0.17exp(−t/77.4) + 0.75  (2.2)  Where t is time and I is the integrated fluorescence intensity over the measured wavelength range. This double decay dynamics is similar to what was observed previously for 442 nm blue light excited skin autofluorescence photobleaching [159]. No significant change was observed in terms of the clinical appearance of the exposed skin area immediately after the experiment or days afterwards. One practical limitation with clinical Raman measurements is that the overwhelming autofluorescence background at some body sites allows only less than 1 s detector integration time, resulting in poor SNR Raman spectra. For example, on the forehead and nose, the fluorescence is very often saturated at 1 s integration time, which does not permit reliable Raman spectrum acquisition. Therefore, 0.3-0.5 s is usually used for these body sites to avoid saturating the spectrometer even though, the quality of Raman spectra is impaired to a large extent. To improve this situation, the idea of using photobleaching was conceived to facilitate longer Raman detector integration time based on the fact that the Raman spectra stay constant during fluorescence photobleaching. As shown in Figure 2.15, with 1 s detector integration time, the spectrum (top curve) was found to be saturated in the range of 400-1000 cm−1 , and therefore, no Raman signals can be extracted from this region, which typically includes many important Raman peaks. Following a 200 s continuous laser exposure photobleaching of the skin, an unsaturated spectrum (bottom curve) with 1 s detector integration time and a very good SNR can be acquired. 49  Figure 2.14: Example data from a female volunteer inner forearm in vivo measurements with 1 s detector integration time. (a) Raw spectra at different laser exposure time points: 1 s, 11 s, 21 s, 51 s, 101 s, and 191 s; (b) extracted fluorescence spectra; (c) extracted Raman spectra; (d) the integrated fluorescence intensity decay curve. In another experiment, by taking advantage of proper photobleaching (power density: 0.29 W /cm2 ; photobleaching time: 200 s), decent Raman spectra with a 2 s detector integration time can be acquired, and the SNR is much better than with a 0.5 s detector integration time, which is illustrated in Figure 2.16(a) and Figure 2.16(b). Five-point smoothing is often used to process the obtained spectra. Figure 2.16(c) and Figure 2.16(d) are the results of smoothing the spectral curves in Figure 2.16(a) and Figure 2.16(b) respectively. Comparing these two curves suggests that smoothing the very noisy Raman spectrum obtained with 0.5 s detector integration time have introduced many artificial Raman peaks, espe-  50  Figure 2.15: In vivo human volunteer forehead skin fluorescence and Raman measurements taken at 1 s (top curve) and at 200 s (bottom curve) after laser exposure with 1 s detector integration time. cially in 500-800 cm−1 , which usually contain rich and important biochemical information from skin.  2.5.3 Conclusion Under continuous exposure of the 785 nm Raman excitation light, in vivo skin autofluorescence signal decays double exponentially with time, while the Raman spectra are not affected by the photobleaching. Pre-exposure of highly fluorescing body sites, such as the forehead and nose, can be used to improve the SNR of clinical Raman spectroscopy. Such photobleaching exposure may be achieved by a high power LED light source that can be conveniently attached to and photobleach the desired skin area before Raman measurements.  51  Figure 2.16: In vivo human volunteer forehead skin Raman measurements taken with (a) 0.5 s and (b) 2 s detector integration time. (c) and (d) 5-point smoothing results of the spectral curves in (a) and (b) respectively.  2.6 Conclusion In summary, a confocal Raman spectroscopy system for skin evaluation was successfully developed. A lateral resolution of 2.2 μ m and a axial resolution of 7.6 μ m were achieved. A study of 494 Raman spectra from 24 mice revealed different spectral patterns at different depths and between normal and tumor-bearing skin sites. A peak at 899 cm −1 (possibly from proline or fatty acids) and one with higher intensity in the 1325-1330 cm −1 range (assigned to nucleic acids) were found in the tumor group. Spectral diagnosis performed on the murine tumor model achieved a diagnostic sensitivity of 95.8% and specificity of 93.8%. These results encourage us to develop the use of confocal Raman spectroscopy  52  further as a clinical tool for noninvasive human skin biochemical analysis, particularly in relation to skin cancer. A method for improving Raman spectroscopy signal-to-noise ratio (SNR) based on fluorescence photobleaching was also presented. A good SNR is essential to obtain biochemical information about biological tissues. Subtracting high levels of tissue autofluorescence background is a major challenge in extracting weak Raman signals. Pre-exposure to laser light was found to significantly reduces tissue autofluorescence, but minimally affects Raman signals, allowing subsequent acquisition of high-SNR Raman spectra. This method was demonstrated with both confocal Raman spectral measurements on ex vivo mouse skin and macro-Raman spectral measurements on in vivo human skin. This method will benefit clinical skin Raman measurements of body sites with high autofluorescence background such as the forehead and nose, and could potentially be used for improving confocal Raman spectral quality as well.  53  Chapter 3 In Vivo Confocal Raman Spectroscopy with Imaging Guidance 3.1 Introduction Confocal Raman spectroscopy provides optical sectioning capability and higher resolution, thereby allowing for more precise in vivo human skin measurements. However, proper interpretation of Raman spectra acquired from the skin requires imaging guidance due to the complexity of skin’s relatively complex stratified architecture. Moreover, within such a small measurement volume (usually in μ m 3 level), any involuntary subject movement including vascular pulsations will affect the acquired Raman spectrum. Therefore, to visualize the target site for the spectral measurement, a reflectance confocal imaging module, which shares the same excitation light source and part of the optics with the developed confocal Raman spectroscopy system will be added into the existing system. The hypothesis of this chapter is that adding a reflectance confocal microscopy (RCM) module will allow us to acquire Raman spectra from specific structures by accurate localization. The hypothesis will be tested by acquiring reflectance confocal images and imaging guided confocal Raman spectra from excised human skin samples and in vivo human skin of volunteer subjects. 54  3.2 System design and development The updated confocal Raman spectroscopy and imaging system (Figure 3.1) was based on the previously described confocal RS system (in Chapter 2) [163] with several improvements described here. Briefly, the laser wavelength was 785 nm. A half waveplate was first introduced to rotate the linearly polarized laser beam and to make sure the power after the the polarization beamsplitter (PBS)(PBS102, Thorlabs, NJ, USA) is maximized. The laser beam then became circularly polarized after passing through a quarter waveplate. An 8 kHz resonance scanner and a galvanometer scanner were mounted closely together to raster scan the laser beam (CRS8 and VM500+, Cambridge Technology, Lexington, MA, USA). The microscope objective was a water immersion objective (LUMPL60×W/IR, numerical aperture (N.A.) 1.0, working distance (W.D.) 2 mm, Olympus, Markham, ON, Canada). This microscope objective has a higher N.A. as compared to the microscope objective described in Chapter 2, thereby generating better optical resolution for the system. The raw spectral signal, which was composed of Raman scattering and the background autofluorescence, was reflected by the same objective and transmitted to the spectrometer through a 50 μ m core-diameter low-OH (low hydroxyl content) single fiber with a high near infrared (NIR) transmission. The spectrometer system was equipped with an NIR-optimized back-illumination deep-depletion charge couple device (CCD) array (Spec10:100BR/LN, Princeton Instruments, Trenton, NJ, USA) and a transmissive imaging spectrograph (HoloSpec- f /2.2-NIR, Kaiser, Ann Arbor, MI, USA) with a volume phase technology holographic grating (HSG-785-LF, Kaiser). The CCD had a 16-bit dynamic range and was cooled with liquid nitrogen to −120 ◦C. Compared with the system described in Chapter 2, the fiber size in this design was changed from 100 μ m to 50 μ m in order to improve the spectrometer resolution of the system from 8 cm −1 to 4 cm−1 . The lateral and axial spatial resolution of the micro-Raman system were measured to be 2 μ m and 6 μ m, respectively. The de-scanned reflected confocal signals were directed by the PBS and then focused by a lens to a 50 μ m pinhole placed in front of the confocal imaging PMT (R3896, 55  Hamamatsu Corp., Bridgewater, NJ, USA). The PMT signals were recorded by an 8-bit multichannel frame grabber (Bitflow Alta, Woburn, MA, USA) to generate the confocal image. For the confocal imaging system, the same laser source was used for excitation. An 8 kHz resonance scanner and a galvanometer scanner were mounted closely together to raster scan the laser beam (CRS8 and VM500+, Cambridge Technology, Lexington, MA, USA). The de-scanned reflected confocal signals were split by a polarization beamsplitter (PBS102, Thorlabs, Newton, NJ, USA) and then focused by a lens. A 50 μ m pinhole was used in front of the confocal imaging PMT (R3896, Hamamatsu Corp., Bridgewater, NJ, USA). The confocal imaging PMT signals were recorded by an 8-bit multichannel frame grabber (Bitflow Alta, Woburn, MA, USA). M  M  L  L  M  l/4 WP  M Scanner  Microscope Objective  M  M  Dichroic  PBS  M M l/2 WP BP  L L  L L  785nm CW laser  LP  Raman Spectometer Fiber holder  Pinhole  Optic fiber PMT  Figure 3.1: Schematic configuration of the system for confocal imaging guided skin Raman measurements. L: lens; BP: bandpass filter; WP: waveplate; M: mirror; PBS: polarization beamsplitter; PMT: photomultiplier tube; LP: long pass filter. 56  (d)  (e)  (f)  Figure 3.2: Raman spectra and reflectance confocal images of ex vivo human skin. Raman spectrum of epidermis (a), dermis (b), fat (c) were acquired from the area shown in the reflectance confocal images: (d) epidermis, (e) dermis, and (f) fat. The image FOV is 300μ m×300μ m. The acquisition time for each Raman spectrum is 10 s.  3.3 Measurements on ex vivo skin samples To test the performance of the developed system, measurements on excised ex vivo human skin from the tip of the nose were conducted. Epidermis, dermis and the fat layers were measured with both Raman spectroscopy and reflectance confocal imaging. Reflectance confocal imaging was first utilized to locate the region-of-interest (ROI), and the scanner was then held stationary to generate still beam illumination for Raman ac-  57  quisition. The scanner was programmed to stop at the center of each image. All the Raman spectra were acquired from the central area (∼2μ m×2μ m) with an integration time of 15s, and the field-of-view (FOV) for all the images was 300μ m×300μ m. From the reflectance confocal image of epidermis shown in Figure 3.2(d), many cellular structures can be seen. The corresponding Raman spectrum is shown in Figure 3.2(a). Figure 3.2(e) shows nice dermal fibers, and strong collagen peaks in the corresponding Raman spectra can be found in Figure 3.2(b). Raman signals of fat are very strong yielding Raman spectrum with good signal-to-noise ratio (SNR) shown in Figure 3.2(c), and the corresponding reflectance confocal image of fat cells can be found in Figure 3.2(f).  3.4 Measurements on in vivo human skin In vivo skin was also measured using the confocal Raman system. The reflectance confocal images were acquired at X-Y plane, which is parallel to the skin surface. A micrometer was used to axially adjust the microscope in order to control depth of imaging. Figure 3.3(a) shows the confocal Raman spectrum from the palm of a normal volunteer, and its corresponding reflectance confocal image is shown in Figure 3.3(d). The stratum corneum is well visualized in the confocal image, and ceramide peaks at 1061 cm −1 , 1128 cm−1 and 1296 cm−1 can be found in the Raman spectrum. A peak at 900 cm−1 was also found, which is close to the peak identified in the tumor mouse skin (Chapter 2). This finding confirmed that this peak is very likely to be associated with keratinization process. In human, the peak at 900 cm−1 is commonly seen when measuring palm skin, which has a thickened stratum corneum layer. However, mouse skin has a very thin stratum corneum layer, therefore, this peak is not supposed to be seen in normal mouse skin. As shown the previous chapter, this peak was only shown in the tumor mouse epidermis, which may indicate an abnormal hyperkeratinization process along with the growth of squamous cell carcinoma in mouse skin. Figure 3.3(b) shows the confocal Raman spectrum of epidermis from the dorsal hand of a volunteer, where strong peaks from DNA can be seen. This is consistent with the  58  reflectance confocal image shown in Figure 3.3(e), where a large number of cells can be found. A junctional nevus from a volunteer was also measured with the developed system. Figure 3.3(c) shows the Raman spectrum and its corresponding confocal image is shown in Figure 3.3(f). Strong melanin fluorescence makes the spectrum noisy, and many bright melanin containing cells are seen in the confocal image (Figure 3.3(f)).  (a)  (d )  (b )  (e )  (c )  (f)  Figure 3.3: Raman spectra and reflectance confocal images of in vivo human skin. Raman spectrum of (a) palm skin, (b) dorsal forearm skin, and (c) junctional nevus were acquired from the area shown in the reflectance confocal images: (d) palm skin, (e) dorsal forearm, and (f) junctional nevus. The image FOV is 200μ m×200μ m. The acquisition time for each Raman spectrum is 10 s.  59  To demonstrate the capability of acquiring confocal Raman spectra from in vivo skin with real-time reflectance confocal microscopic imaging guidance (without holding the scanner stationary), palm skin of a 26-year-old female volunteer was measured. For this measurement, the confocal Raman spectrum was taken with the beam scanning for an area of 200μ m×200μ m. Both the Raman spectrum and the representative reflectance confocal image are shown in Figure 3.4. The major Raman peaks from this spectrum are consistent with previous published results [69].  Figure 3.4: Raman spectra and reflectance confocal images of in vivo human palm skin (with scanning). Image FOV = 300μ m×300μ m. The acquisition time for the Raman spectrum is 10 s.  3.5 Discussion and conclusion There are a few reasons of using a polarization beamsplitter to direct the reflectance confocal signals in this system. One consideration is the relatively low output power (usually less than 100 mw) from the currently existing single-mode continuous wave lasers. Most of the confocal Raman spectroscopy and imaging systems employed two separate lasers 60  for Raman spectral measurement and reflectance confocal imaging [101]. Due to wavelength, beam property, and beam registration variations of the two lasers, this configuration cannot guarantee that the Raman spectral measurement and confocal image are from the same area at the micron-scale level. Therefore, the quality of accurate imaging guidance may be compromised. To avoid this problem, single laser with a 50/50 beamsplitter (BS) was used in some of the confocal Raman systems [164]. However, the illumination will be immediately reduced by half after passing through the BS. Therefore, with further loss caused by other optical components, the power after the microscope objective may not be enough to get confocal Raman spectrum with high signal-to-noise ratio. By using a polarization beamsplitter, the illumination power ideally can 100% pass through. In the developed system, an illumination power of 25 mw at the skin can be achieved. Another consideration is the collection efficiency of the reflectance confocal signals. In systems that use a 50/50 beamsplitter, only 50% of the signals can be reflected. In comparison, in our system, when the circularly polarized signals are reflected from the sample, they will become linearly polarized after passing through the quarter waveplate one more time. The reflectance confocal signals, which are orthogonal to the polarization of the illumination light, will be collected by the PBS and directed to the detector. Another advantage of using the combination of a PBS and a quarter waveplate is that the surface reflection from any optics before the quarter waveplate is rejected. One flaw with the PBS configuration is that when the excitation beam is focused deep enough into the tissue, the polarization of the scattered light signals is likely to be scrambled, resulting a decreased collection efficiency compared to the more superficial cases. But the collection efficiency should still be ≥ 50%, which is better compared to the system using a 50/50 beamsplitter. In this chapter, a reflectance confocal microscopy system was integrated with the confocal Raman spectroscopy system. With the reflectance confocal microscopy added, better visualization to the sample is achieved. The imaging guidance from the reflectance confocal microscopy also provides the capability of measuring specific structures of the sample,  61  which may be critical when the illumination spot size is smaller than the measured sample. Most of the previous studies using confocal Raman spectroscopy to study cells, had no imaging guidance [165, 166]. Measurements on both ex vivo and in vivo human skin were also performed, and high quality reflectance confocal images and confocal Raman spectra were obtained. It is important to demonstrate the capability of monitoring confocal Raman spectral measurements with the confocal imaging module, which could help monitor the degree of movement from the volunteer during the spectral measurement. It is believed that with the morphological information and guidance from the reflectance confocal microscopic module, confocal Raman spectroscopy and imaging system may prove to be not only helpful in early diagnosis of skin cancer and other skin diseases, but also useful in tumor margin assessment and microscopically evaluating treatment effects for a range of skin disorders. Moreover, the successful completion of the described system also paves the road for the next-step experiments of the thesis study, in that the scanning part of the developed reflectance confocal imaging system can be incorporated into the proposed multiphoton microscopic imaging system.  62  Chapter 4 In Vivo Video Rate Multiphoton Microscopy A version of Sections 4.1, 4.2, and 4.5 in this chapter has been published: Lee, A.M.D, Wang, H., Yu, Y., Tang, S., Zhao, J., Lui, H., McLean, D.I., Zeng, H. (2011), In vivo video rate multiphoton microscopy imaging of human skin. Optics Letters, 36(15): 2865-2867. Edits have been made to better integrate this publication into the flow of the thesis.  4.1 Introduction As discussed in the previous chapter, the spatial filters used in confocal microscopy provide optical sectioning capability of thick tissues and cells [167]. An alternative approach to achieve non-invasive optical sectioning is multiphoton microscopy. In reflectance confocal microscopy, refractive index differences provide the optical signal source, whereas in multiphoton microscopy, an image arises from two photon fluorescence (TPF) and/or second harmonic generation (SHG) signals. Previous studies have shown that TPF of cells mainly originates from the cytoplasm, and the major fluorophore in the cytoplasm is reduced nicotinamide adenine dinucleotide (phosphate) (NAD(P)H) [168]. Elastic fibers in the dermis also have strong TPF signals [169]. Additional endogenous fluorescent biomolecules  63  in human skin include keratin, melanin, flavin, porphyrin, tryptophan, cholecalciferol, and lipofuscin [169]. Certain biological materials can be highly polarizable and assemble into large, ordered noncentrosymmetric stuctures, thereby generating SHG signals [170]. Examples include collagen [171, 172], microtubules [173], and myosin [174]. So far, collagen is the only molecule within the skin that has been reported to be active in generating SHG [171, 172]. Collagen and elastic fibers inside dermis can therefore be distinguished via their different optical behaviors under multiphoton excitation. Type I collagen has been identified to be the major signal source of skin SHG [175]. SHG imaging has also proved to be able to discriminate between type I and type III collagen in skin [176]. Therefore, with the capability of acquiring TPF and SHG images simultaneously, multiphoton microscopy can provide both morphological and structural information of cells and extracellular matrix of the skin [113, 177]. One significant limitation to the clinical application of existing multiphoton microscopy (MPM) technologies is the time required to capture each image frame. Imaging speed is important for two reasons: to decrease blurring effects due to subject movement and to image large or multiple skin lesions in a practical time frame. Current in vivo MPM instruments have varying imaging rates from 0.04-2 frames per second (fps) (0.5-24 s/ f rame) [112, 116–118, 178]. The frame rate is limited by the speed of the scanning mechanism, the intensities of the multiphoton signals generated, and the upper limit of the excitation power level acceptable for in vivo imaging. The aforementioned MPM systems primarily use galvanometer scanning mechanisms and photocounting detection. For faster frame rates, rapid scanning mechanisms such as resonance scanners [179] and spinning polygon scanners [180] have been used for ex vivo MPM imaging, but none have yet been employed for in vivo exogenous contrast agent-free human imaging. One way of overcoming the slow imaging speed of multiphoton microscopy (MPM) for clinical application is to design and construct a resonance-scanner-based in vivo MPM instrument. The hypothesis is that such an improved MPM system should be able to capture  64  human skin images at video frame rates and provide good image quality. An additional goal is to perform volumetric measurements of in vivo human skin.  4.2 System design and development The multiphoton microscope setup is shown schematically in Figure 4.1. The output from a tunable (720-950 nm) 90MHz femtosecond Ti:Sapphire laser (Chameleon, Coherent Inc., Santa Clara, California) was expanded with a telescope to match the optical aperture of the optical scanners. An 8 kHz resonance scanner and a galvanometer scanner were mounted as close as possible together to raster scan the laser beam (CRS8 and VM500+, Cambridge Technology, Lexington, Massachusetts). A data acquisition (DAQ) device (USB-6353, National Instruments, Austin, TX, USA) was used to control timing and synchronization of different components. Acquiring images with 256 and 512 lines could generate a maximum frame rate of (7916/256 =) 31 f ps and (7916/256 =) 16 f ps, respectively. The galvanometer flyback time reduced the working frame rates to approximately 27 f ps and 15 f ps, respectively. Beyond the optical scanners, a scan and tube lens combination was used to further expand the beam to fill the rear aperture of a 60× (N.A.=1.0) long working distance (∼2mm) water immersion objective (LUMPLFLN60X/W, Olympus Canada, Markham, Ontario). Adjusting the scanning angles of the resonance and galvanometer scanners (up to 15◦ optical) permitted a variable field of view (FOV) up to 300μ m × 300μ m. A 665 nm excitation long pass dichroic (F665Di02,25×36, Semrock, Rochester, NY, USA) was used immediately behind the objective to reflect the epi-directed multiphoton signals towards the detection arm. The objective and detection arms were placed at the end of a cantilevered arm close to the edge of the optical table to allow accessibility to various skin sites. The DAQ device is the central component of the scanning system. It outputs a waveform to the scanning mirrors, and also sends synchronization signals to the frame grabber. All of the DAQ’s functions are clocked to the resonance scanner’s 8 kHz frequency gener-  65  Figure 4.1: In vivo video rate MPM setup. The SHG/TPF dichroic and filters preceding PMTs were changed or removed according to the imaging modalities. (a) Example SHG+TPF image extracted from a video (27 f ps) of the dorsal forearm of a 24-year-old Asian female (Supplementary Media 1). Excitation wavelength = 730 nm. Both TPF and SHG photons were directed onto a single PMT. FOV = 200μ m×200μ m. Resolution = 256pixels×256pixels. Frame scan rate = 27 f ps. Frame capture/display rate = 24 f ps. ated as a digital pulse each time it completes an oscillation and returns to the left side of a frame. This is the frame horizontal synchronization (H. Sync signal). The DAQ uses the H. Sync to implement a counter which outputs a vertical synchronization (V. Sync signal) signal after a certain number of H. Sync signals. The frame grabber uses the H. Sync pulse redirected from the DAQ to time the beginning of each new imaging line, and the V. Sync to time the beginning of each frame. After each V. Sync, the frame grabber will read either 256 or 512 imaging lines, depending on the user specification. In the detection arm, a 710 nm short pass emission filter was used to remove residual excitation light. Depending on the excitation wavelength, selected dichroic mirrors and filters were used to separate and direct the SHG and TPF signals to a pair of photomultiplier tube (PMT) modules (active area 3.7mm×13mm, > 15% quantum efficiency between 200 nm and 650 nm, H9433MOD−03, Hamamatsu Corp., Bridgewater, New Jersey). The small size of the PMT modules (19mm×53mm×51mm) allowed for construction of a compact 66  detection arm with close proximity of the PMTs to the objective to maximize the SHG and TPF collection efficiencies. Previously reported systems commonly use PMTs in photon counting or digital mode to detect the low TPF and SHG signal levels by counting single photons. As the upper count limit for a photon counting unit is typically of the order of megahertz, this necessitates longer image acquisition times. The PMTs in our system were operated in analog mode to handle the signals at the faster speed needed for video rate imaging. The SHG and TPF PMT signals were recorded by an 8-bit multichannel frame grabber (Bitflow Alta, Woburn, Massachusetts) as synchronized but separate video streams. As the resonance scanner operates bidirectionally, the forward and backward passes of each line were added during postprocessing. To further improve image quality, the acquired images were rebinned accordingly to yield a frame size of 256pixels×256pixels.  4.3 Imaging extracted biological samples Extracted biological samples, including bovine collagen and bovine elastin were first measured to test the system performance. Figure 4.2(a) shows the two photon fluorescence (TPF) image of the bovine elastin sample, where short and fine fibers can be seen. Second harmonic generation (SHG) image of bovine collagen sample shown as thick fiber bundles and fine fibril structures can be found in Figure 4.2(b).  4.4 Imaging ex vivo human skin Ex vivo human skin samples were also imaged. Figure 4.3(a) shows TPF image of elastin at dermis and Figure 4.3(b) shows SHG image of collagen at the same layer.  4.5 Imaging in vivo human skin In vivo human skin was then studied with the multiphoton microscopic system. To facilitate effective optimization of excitation wavelengths for imaging different skin structures, the system was configured into a special integrated SHG and TPF imaging mode (denot-  67  (b )  (a)  Figure 4.2: Multiphoton fluorescence and second-harmonic-generation images of extracted (a) bovine elastin and (b) bovine collagen. Excitation wavelength = 800 nm, field of view = 150μ m×150μ m.  (a)  (b )  Figure 4.3: Two photon fluorescence (TPF) and second-harmonic-generation (SHG) images of an ex vivo human skin sample: (a) TPF image of dermal elastin; (b) SHG image of dermal collagen. Excitation wavelength =800 nm, field of view = 150μ m×150μ m.  68  ed as SHG+TPF) by removing the SHG/TPF dichroic and the filter in front of the PMT. This way, both the SHG and the TPF photons were directed onto a single PMT. By varying excitation wavelengths, the SHG+TPF images can be obtained conveniently without the need to change the SHG+TPF dichroic or filters. This new imaging modality is very useful for excitation wavelength optimization in a clinical setting, where fast measurement is essential. For more stable imaging, a metal ring (∼ 3 cm in diameter) was fixed to the skin of volunteer subjects using a double-sided adhesive film. No coverslip was present between the objective and the skin surface. Water was placed within the ring prior to coupling it with a magnetic holder affixed to the instrument. The holder was mounted to a manually actuated three dimensional translation stage to control the imaging location and depth. The laser power incident on the skin at all wavelengths was adjusted to 40 mW using a half-waveplate/polarizer combination at the laser exit. The study was approved by the University of British Columbia Research Ethics Board (#H96 − 70499). Informed consent was obtained from each volunteer subject. Images were acquired from the dorsal forearms of volunteer subjects. It was found that the optimal excitation wavelength for imaging the cellular layers in the epidermis was 730 nm, which is consistent with the excitation maxima of fluorophores within the epidermis such as keratin and NAD(P)H [178]. Dermal structures such as collagen and elastin are best imaged at 800 nm excitation. The image in Figure 4.1(a) was acquired in 0.036 s while those shown in Figure 4.4Figure 4.6 were acquired in 0.067 s. The FOV for these images is 200μ m×200μ m. Figure 4.1(a) shows an integrated SHG+TPF image from epidermis under 730 nm excitation. TPF images of epidermis under 730 nm excitation are shown in Figure 4.4. Images near the skin surface (Figure 4.4(a)-(c)) show a somewhat amorphous stratum corneum (SC) overlaying the larger cells of the stratum granulosum (SG) and the stratum spinosum (SS), while smaller cells of the stratum basale (SB) are present at the epidermal/dermal boundary (Figure 4.4(d)-Figure 4.4(f)). In the SG and SS, the cell nuclei are dark while the  69  cytoplasm is fluorescent. The SB cells appear relatively bright, presumably due to melanin fluorescence. Figure 4.5 shows integrated SHG+TPF images under 900 nm excitation. The papillary structure at the epidermal/dermal boundary is clearly visible. Although cells in the upper epidermis are not visualized at this wavelength [178], the basal cells are clearly seen, as are the fibrous structures in the dermis. False color overlay images of TPF and SHG acquired using 880 nm excitation are shown in Figure 4.6. The images collected from the reticular dermis show elastic fibers in the TPF channel and collagen fiber bundles in the SHG channel. The collagen fibers shown in Figure 4.6(a) are long and bundle-shaped, whereas Figure 4.6(b) shows short and amorphous collagen fibers, which may indicate collagen degradation due to photoaging or chronological aging [181]. The images (Figure 4.4 − 4.6) and videos (Supplementary Media 1-4) presented here show some distortion near the left and right edges because they have not been corrected for the sinusoidal scan pattern of the resonance scanner. Distortion on the left side of the images is mitigated by delaying the start of the line acquisition relative to the resonance scanner turning point on the left side. The corresponding videos where Figure 4.1(a) and Figure 4.4 − 4.6 were generated are shown in Supplementary Media 1, 2, 3, and 4, respectively. The frames of Supplementary Media 1 were captured with 256 lines, yielding a scan frame rate of 27 f ps (frame acquisition time = 0.036 s). Latency in the frame grabber used in the system resulted in dropped frames that reduced the captured frame rate to 24 f ps. Utilizing a different frame grabber should readily capture the full frame rate. Supplementary Media 2 − 4 were captured with 512 lines, yielding a frame scan rate of 15 f ps (frame acquisition time = 0.067 s, captured frame rate = 12 f ps). The image quality improvement when scanning at the slower frame rate is noticeable as expected with the doubled acquisition time per frame. Clearly evident in all of the videos are the frame-to-frame jumps in the Z imaging plane due to subject motion, thus emphasizing the need for fast imaging frame rates. Even when the subject motion is consciously reduced, hemodynamic pulsations can be seen in some instances. Imaging  70  Figure 4.4: TPF images from epidermis (oblique orientation) extracted from an in vivo video from the dorsal forearm of a 25-year-old Asian female increasing in depth from (a) near the surface to (f) near the dermal/epidermal boundary (Supplementary Media 2). A 405 nm dichroic mirror ensured only TPF signals reached the PMT. Excitation wavelength = 730 nm. FOV = 200μ m×200μ m. Resolution = 256pixels×256pixels. Frame scan rate = 15 f ps. Frame capture/display rate = 12 f ps. At this wavelength, the cells of the epidermis are well visualized from the SC down to the SB. Scale bar: 20 μ m. at a slower frame rate could result in significant blurring of the image. In summary, it is believed that the high frame rate imaging capabilities provided by the optimized in vivo MPM instrument and the new integrated SHG+TPF imaging modality are unique amongst currently available systems and are necessary embodiments to the practical use of MPM in a clinical setting.  4.6 Volumetric imaging demonstrating optical biopsy The multiphoton system was further modified to provide the capability of three dimensional (3D) imaging, which also demonstrates the concepts of 3D optical biopsy. Instrumentally, a new control over the Z-movement of the microscope objective lens 71  Figure 4.5: Integrated SHG+TPF images of the epidermal ridges and the papillary dermis extracted from an in vivo video of the dorsal forearm of a 28-year-old Asian male (Supplementary Media 3). The imaging depth increases from (a) to (f) starting from the dermal/epidermal boundary to the papillary dermis. Excitation wavelength = 900 nm. TPF and SHG photons were directed onto a single PMT. FOV = 200μ m×200μ m. Resolution = 256pixels×256pixels. Frame scan rate = 15 f ps. Frame capture/display rate = 12 f ps. At this wavelength, the cells of the SB and collagen and elastin fibers are observed. Scale bar: 20 μ m. was introduced by a piezo-electric actuator (MIPOS−500, Piezosystem-Jena, Hopedale, MA, USA). This actuator comes with an amplifier controller, which can be modulated using an external voltage signal to move the actuator its full range of 400 μ m. By modification to the software, it is possible to achieve two distinct scanning modes: XY-plane scanning, and XZ-plane scanning. XY-plane scanning is the more common method of imaging, but XZ-scanning emulates conventional sectioning done in histology and is a desirable functionality of the imaging system. For XY-plane scanning, the image frames are in the XY-plane, and stacks of images can be taken in the Z-direction using the piezo actuator shown in Figure 4.7(a). Similarly, for XZ-plane scanning, the piezo-actuator moves  72  Figure 4.6: False color overlay of SHG (green) and TPF (red) images from two different areas ((a) and (b)) of the reticular dermis extracted from an in vivo video of the dorsal forearm of a 63-year-old Caucasian male (Supplementary Media 4). A 458 nm dichroic mirror was used to separate the SHG and TPF signals. 440/40 nm bandpass and 458 nm long pass filters were placed in front of the SHG and TPF PMTs, respectively. Excitation wavelength = 880 nm. FOV = 200μ m×200μ m. Resolution = 256pixels×256pixels. Frame scan rate = 15 f ps. Frame capture/display rate = 12 f ps. Collagen (SHG) and elastin (TPF) fibers are clearly seen. Scale bar: 20 μ m.  (b )  (a)  Figure 4.7: Illustration of stacks of images in (a) XY-plane and (b) XZ-plane. within each frame, which allows the image frames in the XZ-plane, and stacks of images can be taken in the Y-direction shown in Figure 4.7(b). In both modes, the resonance scanner is controlled with a single analog voltage level, and the horizontal field of view (FOV) is increased or decreased by changing this voltage level.  73  Figure 4.8: 3D reconstruction of collagen fibers from XY-Z scan on left and XZ-Y scan on right.  4.6.1 Volumetric imaging on extracted biological samples Extracted human collagen was first imaged using XY-scanning mode. The step size in Z-direction was set to be 6 μ m, and the FOV of XY imaging plane was 150μ m×150μ m. The 3D image can be found in Figure 4.8(left). Human collagen sample was then measured using XZ-scanning mode with a stepsize of 6 μ m in Y-direction and a XZ FOV 150μ m×150μ m. The excitation laser wavelength was 800 nm, the power at the sample was 20 mW . The total acquisition time was 150 s. The 3D image can be found in Figure 4.8(right). All the 3D reconstructions were created using a modeling program named Amira (Amira, Visage Imaging, Inc., San Diego, CA, USA). The 3D models looked identical to one another, verifying that the DAQ properly coordinates the scanning mirrors and piezo-actuator. The model from the XZ-Y scan is mirrored because the imaging software renders from the bottom of the data to the top, while the galvanometer scans from the top of a stack to the bottom.  4.6.2 Volumetric imaging on ex vivo human skin Ex vivo human skin excised from the cheek was also measured using XY-scanning mode. The stepsize in Z-direction was set to be 0.5 μ m, and the FOV of XY imaging plane was  74  Figure 4.9: 3D reconstruction of ex vivo human cheek skin from XY-Z scan (Supplementary Media 5). Acquisition time: 2 min; depth: 50 μ m; stepsize: 0.5 μ m; FOV: 150μ m×150μ m; laser wavelength: 730 nm; and laser power: 30 mW . 150μ m×150μ m. The excitation laser wavelength was 730 nm, and the power at the sample was 30 mW . To improve the reconstructed 3D image quality, 10 scans per depth were performed with a total acquisition time of 2 min. The reconstructed 3D stack, which was averaged over the 10 scans, can be visualized in Supplementary Media 5, where slices of skin showing individual cellular structures from the superficial surface to the 50 μ m beneath the surface are seen. A representative 3D image extracted from Supplementary Media 5 can be found in Figure 4.9. Similarly, an ex vivo human skin sample excised from the eyelid was measured using the XY-scanning mode. The stepsize in Z-direction was set to be 1 μ m, and the FOV of XY imaging plane was 150μ m×150μ m. The excitation laser wavelength was 730 nm, the power at the sample was 30 mW . The total acquisition time was 2 min. The first part of Supplementary Media 6 shows the rotational views of the reconstructed 3D stack, and the second part of the media shows individual slices of skin from the superficial surface to the 130 μ m depth. Epidermal keratinocytes, bright basal cells, and dermal fiber structures can be well visualized. A representative 3D image extracted from Supplementary Media 6  75  Figure 4.10: 3D reconstruction of ex vivo human eyelid skin from XY-Z scan (Supplementary Media 6). Acquisition time: 2 min; depth: 130 μ m; stepsize: 1 μ m; FOV: 150μ m×150μ m; laser wavelength: 730 nm; and laser power: 30 mW . showing both basal cells and dermal fibers is shown in Figure 4.10.  4.6.3 Volumetric imaging on in vivo human skin More tests were carried out with in vivo human skin imaging to obtain XY-Z stacks of images. Figure 4.11 shows the reconstruction of a sweat gland from a volunteer. The stepsize in Z-direction was set to be 1 μ m, and the FOV of XY imaging plane was 150μ m×150μ m. The excitation laser wavelength was 730 nm, the power at the sample was 30 mW . Ten scans per depth with a total acquisition time of 2.5 min was also performed. However, due to movement from the volunteers, averaging over 10 scans will make the image blurry. Therefore, single scan data was used to reconstruct the 3D stack, which is shown in Supplementary Media 7 with a representative image shown in Figure 4.11. Cellular structures can be seen, but with a lower SNR comparing with the averaged ex vivo data. It was also possible to locate a sweat duct during the in vivo imaging, which is shown in Supplementary Media 8. The spiral shaped sweat duct can be well visualized in the representative image in Figure 4.12. 76  Figure 4.11: 3D reconstruction of in vivo human forearm skin from XY-Z scan (Supplementary Media 7). Acquisition time: 15 s; depth: 150 μ m; stepsize: 1 μ m; FOV: 150μ m×150μ m; laser wavelength: 730 nm; and laser power: 30 mW .  Figure 4.12: 3D reconstruction of in vivo human sweat duct from XY-Z scan (Supplementary Media 8). Acquisition time: 15 s; depth: 150 μ m; stepsize: 1 μ m; FOV: 150μ m×150μ m; laser wavelength: 730 nm; and laser power: 30 mW .  77  In summary, the addition of a piezo-actuator introduces useful functionality to the multiphoton imaging system. A user can more accurately determine the position of the focal plane, and take stacks of images in the XY plane as well as in the XZ plane, which provide more information compared with one dimensional images and better mimic conventional histological sectioning results.  4.7 Discussion In this chapter, excitation power at the skin was controlled to be below 40 mW . With an illumination spot of 1 μ m in diameter, the power density is estimated to be less than 5×10 6 W /cm2 , which is well below the safety threshold (8×10 11 W /cm2 ) reported in the literature [182]. For wavelength selection, previous studies have shown that wavelengths between 730800 nm are able to excite NAD(P)H two photon fluorescence [183, 184] with an optimal excitation wavelength at 730 nm [178]. Therefore, 730 nm was selected to image epidermis in this study. In contrast, under one-photon excitation, melanin and elastin fluoresce at around ∼550 nm [185], which is much longer than the NADPH emission wavelength (∼450 nm) [186]. To image the dermal-epidermal junction, which mainly consists of melanin-containing basal cells and dermal fiber structures, 900 nm laser wavelength was employed in the study. In order to choose an optical wavelength for reticular dermis imaging, combined SHG+TPF images were first taken on both ex vivo and in vivo skin with different excitation wavelengths from 730 nm to 950 nm. The results showed that the combined SHG+TPF imaging channel generated the optimal signals from human dermis at an excitation wavelength of 880 nm. Therefore, this particular wavelength was selected, and corresponding bandpass and long pass filters were used to separate SHG from TPF signals at dermis level. In Figure 4.4, near the skin surface, the cytoplasm appears as bright with dark cell nucleus in the middle of each cell. The size of the cell decreases with an increased number  78  when imaging deeper into the skin. Bright basal cells are presented at the junctional layer, which indicates a higher amount of melanin presented in those cells. These results are consistent with previously published work [185, 187]. Figure 4.5 shows a horizontal view from the upper tip of the dermal papilla down to the papillary dermis of the skin. At the upper most portion of the dermal papilla shown in Figure 4.5(a) and (b), mainly bright basal cells are seen. Going deeper, both dermal fibers and basal cells are seen as in Figure 4.5(c)-(f). Interestingly, as shown in Figure 4.5(e) and (f), there is a dark circular area present in almost all the papilla. Since fibers should be everywhere in the dermis, a completely filled in bright fiber area is expected with surrounding basal cells as in (d). One possibility is that the dark area may represent a blood vessel or a capillary inside the skin because blood will absorb most of the light and the vessel will appear as dark in the multiphoton images. Yet another finding is that there are yellow structures present in the false color overlaid SHG (green) and TPF (red) images as shown in Figure 4.6(b), which is generated by overlapping from the two imaging channels. One possible explanation is that there are co-locolizations of collagen and elastic fibers. Another possibility is that certain types of collagen are able to generate both TPF and SHG signals which will be shown in both of the two imaging channels. It is also possible that there might be other structures inside the skin dermis which could generate both TPF and SHG signals. To confirm these findings, further histology analysis will be needed. It was found that collagen fibers can generate strong SHG signals. The reason is that SHG (frequency doubling) requires intense laser light passing through a highly polarizable material with a noncentrosymmetric molecular organization. This means that SHG requires an environment lacking an inversion center [170]. Collagen fiber is composed of a dense array of polar collagen fibrils, which are arranged with their long axes nearly parallel. Collagen fibrils are made of collagen molecules, which contain 3 polypeptide chains and super-coiled around the common axis to form a triple helix structure [188]. This unique organization of collagen fibrils provides its macroscopic polarity, which makes no energy  79  loss and allows phase-matching between the fundamental wave and the second-harmonic wave, therefore, has very high SHG yields [189]. One limitation with the current study is that only a few volunteers have been measured. The scope of this work was primarily to solve the technical challenges of developing a MPM microscope system for in vivo work. Experiments on additional volunteers and skin patients will better characterize the system performance and explore the capability of the developed instrument.  4.8 Conclusion A resonance scanner-based multiphoton microscopy instrument specially designed for in vivo dermatological use that is capable of imaging human skin at 27 frames per second with 256pixels×256pixels resolution without the use of exogenous contrast agents has been developed and tested. Imaging at fast frame rates is critical to reducing image blurring due to patient motion and to providing practically short clinical measurement times. High quality second harmonic generation and two-photon fluorescence images and videos acquired at optimized wavelengths were presented showing cellular and tissue structures from the skin surface down to the reticular dermis. Volumetric information of both ex vivo and in vivo human skin, and 3D reconstructions of the skin images can be obtained with the developed MPM system. The fast imaging speed of the developed instrument shortened the acquisition time for volumetric measurement of human skin to be within a few minutes, which may be practical for facilitating clinical applications. The unique functions of taking images in both XY plane and XZ plane may also provide more comprehensive information for better skin evaluation and diagnosis.  80  Chapter 5 Co-registered Multiphoton and Reflectance Confocal Video Rate Imaging of In Vivo Human Skin A version of Sections 5.1, 5.2 and 5.5 in this chapter has been published as: Wang, H., Lee, A.M.D., Frehlick, Z., Lui, H., McLean, D.I., Tang, S., and Zeng, H. (2013), Perfectly registered multiphoton and reflectance confocal video rate imaging of in vivo human skin. Journal of Biophotonics, 6(4): 305-309. Edits have been made to better integrate this publication into the flow of the thesis.  5.1 Introduction The development of non-invasive diagnostic imaging techniques for examining the microscopic structure of skin is important because the standard examination practice of biopsy leads to scarring. Two techniques that have garnered much interest in recent years for dermatology use are reflectance confocal microscopy (RCM) and multiphoton microscopy (MPM). The optical sectioning capability of RCM has allowed in vivo, high resolution morphological imaging of skin [190] non-invasively. Comparing with RCM, MPM also has  81  inherent optical sectioning capability due to the nonlinear excitation process that obviates the need for a pinhole to reject out-of-focus light. Different MPM excitation mechanisms are sensitive to different biochemical components of the tissue. For example, two-photon fluorescence (TPF) signals arise from endogenous fluorophores of skin components such as elastin, nicotinamide adenine dinucleotide phosphate (NAD(P)H), and keratin; while second harmonic generation (SHG) is sensitive to noncentrosymmetric structures such as collagen [119, 171]. The term noncentrosymmetry refers to a point group which lacks of an inversion center as one of its symmetry elements [191]. As there is less scattering and absorption of the near infrared (NIR) lasers used in MPM, the light has a deeper penetration depth and induces less photo-damage to the tissue compared to the equivalent single photon excitation laser [119, 172]. Combining both RCM and MPM images (hereafter called RCM/MPM imaging) potentially allows greater clinical diagnostic utility as complementary information can be revealed using the two techniques. Indeed there have been a number of studies of this nature involving both ex vivo and in vivo tissue imaging. Images of ex vivo porcine skin and bovine cornea using RCM/MPM imaging have shown that cell nuclei can be detected in the reflectance confocal signal, while the multiphoton autofluorescence signal can be used for cytoplasmic imaging through the entire epithelium [192]. Similarly, ex vivo human skin studies have demonstrated that cell borders are more clearly seen from the reflected confocal signal,while the autofluorescence signal provides the cytoplasmic details [193, 194]. In vivo RCM/MPM imaging has also been used together to evaluate skin disease such as seborrheic keratoses, angioma, and actinic keratoses, but without co-registration of the two imaging modes [195]. As for a clinical application, in vivo imaging is preferred over ex vivo imaging because it does not necessitate tissue removal. It also leaves the tissue in its native state, whereas ex vivo tissue can be subject to biochemical/ structural changes due to the degradation of the sample, tissue contraction, and elimination of living tissue dynamics such as blood  82  perfusion and oxygenation. With respect to in vivo skin imaging, the previously reported studies usng RCM/MPM have serious limitations. This type of imaging is difficult because patient motion must be mitigated, and often multiple lesion sites or large lesions must be examined. Various motion compensation approaches such as real-time adaptive focus control [196] have been applied for reducing motion artifacts, suitable for imaging fast dynamics in a single focal plane. Surveying a tissue volume by imaging through many focal planes, while also limiting motion artifacts can be simultaneously handled by imaging with as high frame rate as possible. Video rate imaging has been demonstrated in RCM [89, 197], but not in a combined RCM/MPM system. Another important issue with in vivo RCM/MPM imaging is ensuring image registration between the RCM and MPM channels. If entirely different optical systems are used to measure the RCM and MPM channels separately, image registration is not guaranteed [168, 195]. Systems have been developed that use the same objective for both imaging modalities by multipassing the tissue with two laser sources. However, RCM/MPM image registration is still not ensured as the patient can still move between passes [198–200]. Another system has scanned RCM and MPM with the same objective simultaneously with different wavelengths but image registration can still be problematic as two different scanning mechanisms were used [201]. The problem of image registration can be solved definitively if a single laser source is used and scanning is made with the same mechanism while imaging with both the RCM and MPM channels. This type of system has been developed but has only been demonstrated using ex vivo specimens [192, 194]. Other than the image registration improvement when using a single laser to perform RCM and MPM, it is also advantageous to have a femtosecond RCM (fsRCM) imaging module compared with the continuous-wave RCM (cwRCM) that was described in Chapter 3. First, with the same Ti:Sappire fs laser used in Chapter 4, tunable wavelength from 720 nm to 950 nm can be achieved, while only single wavelength at 785 nm can be used from  83  the cw laser. It has been reported in the literature that both qualitative and quantitative differences were observed between the RCM images acquired from the skin with different excitation wavelengths, and it was concluded that the selection of excitation wavelength is important to properly understand and resolve different biological structures of human skin [202]. Second, since cwRCM will use a different laser from the MPM laser, the cwRCM images cannot be acquired simultaneously with the MPM images. Therefore, in this chapter, the goal is to modify the previous system described in Chapter 4 and develop a fsRCM image module. The hypothesis of this chapter is that a combined fsRCM and MPM system, which could capture co-registered fsRCM and MPM (both SHG and TPF) images at video rate can be achieved. Another hypothesis is that the fsRCM and MPM imaging channels should provide complementary information of the skin. The hypotheses will be tested by acquiring both fsRCM and MPM images of extracted biological samples, ex vivo skin, and in vivo skin using the proposed instrument. For simplicity, the fsRCM will be abbreviated as RCM hereinafter in this chapter.  5.2 System design and development The system setup is shown schematically in Figure 5.1. As it is based on the MPM system that has been described in the previous chapter, details on the system can be found elsewhere [197]. Briefly, the output from a tunable (720-950 nm) femtosecond Ti:Sapphire excitation laser was scanned over the back aperture of a 60× (N.A. = 1.0) water immersion objective using an 8kHz resonance scanner for the fast axis and a galvanometer scanner for the slow axis. The maximum field of view (FOV) for the system was 300μ m × 300μ m. Acquiring images with 256 or 512 lines generates frame rates of 27 f ps or 15 f ps respectively including galvo flyback time. The MPM channels were collected in the epi-direction using a 665 nm longpass dichroic beamsplitter. Two modes were used to collect the MPM signals: one where both the TPF and the SHG signals were integrated into a single PMT (SHG+TPF imaging mode); and the other where insertion of another dichroic beam-  84  Figure 5.1: In vivo video rate RCM/MPM system setup. The SHG+TPF dichroic and filters preceding PMTs were changed or removed according to the desired imaging modalities. Lenses have been omitted from the diagram for clarity. splitter, filters and use of two PMTs allowed separation of the TPF and SHG images. A 50/50 beamsplitter was used to direct the descanned RCM signal through a 50μ m pinhole to an avalanche photodiode (APD) module (C10508, Hamamatsu Corp., Bridgewater, NJ). The RCM, TPF, and SHG images were recorded by a 10-bit multichannel frame grabber (Helios eA, Matrox Electronic Systems Ltd., Dorval, QC, Canada) as synchronized independent video streams. As the resonance scanner operates bidirectionally, the forward and backward passes of each line were added in real-time. Real-time correction of the resonance scanner sinusoidal image distortion was also applied. The pixel clock was matched to the PMT amplifier bandwidth and pixels were re-binned to reduce the frame size to 256×256 pixels. As all the imaging channels share the same laser, scanners, and objective, the RCM, TPF, and SHG images are well co-registered, ensuring with complete certainty that the images are comparable.  85  (a)  (b)  (c)  Figure 5.2: (a) RCM, (b) SHG+TPF, and (c) false color overlay of SHG+TPF (green) and RCM (red) images from extracted bovine collagen sample. Excitation wavelength = 785 nm. FOV = 300μ m × 300μ m. Resolution = 256pixels×256pixels. Frame rate = 15 f ps.  (a)  (b)  (c)  Figure 5.3: (a) RCM, (b) SHG+TPF, and (c) false color overlay of SHG+TPF (green) and RCM (red) images from extracted bovine elastin sample. Excitation wavelength = 785 nm. FOV = 300μ m × 300μ m. Resolution = 256pixels×256pixels. Frame rate = 15 f ps.  5.3 Imaging extracted biological samples Extracted bovine collagen and elastin samples were first measured with both reflectance confocal and the combined SHG and TPF (SHG+TPF) imaging channels. Figure 5.2 shows the RCM, SHG+TPF and the false color overlay of SHG+TPF in green and RCM in red images from bovine collagen sample. Figure 5.3 shows the RCM, SHG+TPF and the false color overlay of SHG+TPF in green and RCM in red images from bovine elastin sample. In both cases, good registrations were found in the RCM and SHG+TPF imaging channels.  86  (a)  (b)  (c)  Figure 5.4: (a) RCM, (b) SHG+TPF, and (c) false color overlay of SHG+TPF (green) and RCM (red) images from epidermis of ex vivo eyebrow skin from surgery. Excitation wavelength = 730 nm. FOV = 100μ m × 100μ m. Resolution = 256pixels×256pixels. Frame rate = 15 f ps.  5.4 Imaging ex vivo human skin Ex vivo eyebrow skin, excised from a 60-year-old male was imaged with RCM and SHG+TPF channels. Figure 5.4(a) shows the RCM image of the eyebrow skin at epidermis level ( 12  μ m beneath the surface of the skin), where honey-comb shaped cellular structures with bright cell boundaries can be seen. Differently, SHG+TPF imaging channel displayed in Figure 5.4(b), cell nuclei appearing in dark and cytoplasm appearing in bright can be clearly seen. Interestingly, it was noticed that there is a bright circular inner boundary between the cell nuclei and the cytoplasm on some of the cells, which could be the double lipid bilayered nuclear membrane. The complementary nature of RCM and MPM can be demonstrated in Figure 5.4(c), which shows the false color overlay of the two imaging channels, where RCM is shown as red, and SHG+TPF as green. Similarly, RCM and SHG+TPF images acquired from the dermis of a piece of ex vivo eyelid skin are shown in Figure 5.5. Overall, dermal fibers are better seen in the combined SHG+TPF imaging channel (Figure 5.5(b)) than in the RCM channel (Figure 5.5(a)). As shown in false color overlaid image (Figure 5.5(c)), the RCM and SHG+TPF images have certain overlaps shown as yellow color, which indicates a high level of both RCM and MPM signals.  87  (a)  (b)  (c)  Figure 5.5: (a) RCM, (b) SHG+TPF, and (c) false color overlay of SHG+TPF (green) and RCM (red) images from dermis of ex vivo eyelid skin from surgery. Excitation wavelength = 750 nm. FOV = 150μ m × 150μ m. Resolution = 256pixels×256pixels. Frame rate = 15 f ps.  5.5 Imaging in vivo human skin To reduce the involuntary subject motion during in vivo measurement, the forearms of volunteer subjects were first attached with a double-sided tape, which on the other side is taped on a metal ring. The metal ring is then magnetically mated with the imaging objective. Only water was present between the objective and the skin surface. The laser power incident on the skin at all wavelengths was adjusted to be equal to 40 mW using a half-waveplate/polarizer combination at the laser exit. The study was approved by the University of British Columbia Research Ethics Board (#H96−70499). Informed consent was obtained from each volunteer subject. Figures 5.6 and 5.7 show the RCM, integrated SHG+TPF, and false color overlay (red = RCM, green = SHG+TPF) skin images using 730 nm excitation. As previously reported [197], shorter wavelengths provide better MPM imaging contrast in epidermis. At shorter excitation wavelengths where the laser penetration is shallow and the MPM signals are only TPF, it is better to image with a single PMT because the detection path length can be made shorter to increase the signal collection efficiency. The RCM (Figure 5.6(a)) and SHG+TPF (Figure 5.6(b)) images near the stratum basale (SB) layer of a Caucasian male in his early 50’s are overlaid in false color in Figure 5.6(c). Images in Figure 5.6 were extracted from Supplementary Media 9, which is a 15 f ps in vivo false color overlay movie composed of  88  Figure 5.6: (a) RCM, (b) SHG+TPF, (c) false color overlay of SHG+TPF (green) and RCM (red), and (d) false color overlay of SHG+TPF (red) and RCM (green) images extracted from Supplementary Media 9 from the stratum basale (SB) of the dorsal forearm of a Caucasian male in his early 50’s. Excitation wavelength = 730 nm. FOV = 170μ m ×170μ m. Resolution = 256pixels×256pixels. Frame rate = 15 f ps. Scale bar: 20 μ m. Black arrows in (c) and (d) indicate cells having both high RCM and SHG+TPF signals. images taken at increasing depths from near the skin surface to near the dermal-epidermal junction (DEJ). The images and video clearly show the complementary nature of the RCM and MPM imaging modalities. The honey-comb shaped cells in the stratum spinosum (SS) show bright cell boundaries in the RCM channel while bright cytoplasm is seen in the SHG+TPF channel. Deeper at the DEJ, dermal papilla and cellular structures are well visualized in both the RCM and SHG+TPF channels. Some basal cells at the DEJ are bright in both the RCM and SHG+TPF imaging channels (yellow in false color overlay, noted by black arrows in Figure 5.7(c) and (d)). It is believed that these cells contain melanin, which  89  is both highly scattering and fluorescent [88, 201]. The bright spot near the center of all of the RCM images and videos is an artifact due to reflection from one of the lenses in the optical system. Figure 5.7 shows the RCM (Figure 5.7(a)), SHG+TPF (Figure 5.7(b)), and false color overlay images (Figure 5.7(c)) from the stratum granulosum (SG) layer of a 41-year-old Asian male. Images in Figure 5.7 were extracted from Supplementary Media 10, which is a 27 f ps false color overlay (RCM=red, SHG+TPF=green) in vivo movie increasing in depth from near the skin surface to near the stratum basale (SB). As expected, the almost doubling of the frame rate in Supplementary Media 10 (Figure 5.7) compared to Supplementary Media 9 (Figure 5.6) results in lower image quality because less pixel averaging leads to a lower signal-to-noise ratio.  (a)  (b)  (c)  Figure 5.7: (a) RCM, (b) SHG+TPF, and (c) false color overlay of SHG+TPF (green) and RCM (red) images extracted from Supplementary Media 10 from the stratum granulosum (SG) of the dorsal forearm of a 41-year-old Asian male. Excitation wavelength = 730 nm. FOV = 150μ m × 150μ m. Resolution = 256pixels×256pixels. Frame rate = 27 f ps. Scale bar: 20 μ m. Figure 5.8 shows RCM/MPM imaging from the ventral forearm of a 64-year-old Caucasian male under 880 nm excitation. Using this longer wavelength permits deeper imaging into the skin dermis. Images from the reticular dermis (RD) are shown in the figure. For these images, a 458 nm dichroic beamsplitter, and 440 nm/40 nm bandpass and 458 nm longpass filters were used in the detection arm to separate the MPM signals onto SHG and TPF PMTs respectively. The simultaneously acquired RCM (Figure 5.8(a)), TPF (Figure 5.8(b)), and SHG (Figure 5.8(c)) images can be viewed separately or overlaid in false color 90  (Figure 5.8(d). The images in Figure 5.8 were extracted from Supplementary Media 11, a 15 f ps false color overlay (RCM=red, TPF=green, SHG=blue) in vivo video varying in depth through the RD.  (a)  (b)  (c)  (d)  Figure 5.8: (a) RCM, (b) TPF, (c) SHG, and (d) false color overlay of RCM (red), TPF (green), and SHG (blue) images extracted from Supplementary Media 11 from the reticular dermis (RD) of the ventral forearm of a 64-year-old Caucasian male. Excitation wavelength = 880 nm. FOV = 150μ m × 150μ m. Resolution = 256pixels×256pixels. Frame rate = 15 f ps. Scale bar: 20 μ m. The RCM image shows structures that appear rather diffuse, while those in the TPF and SHG images are finer and more pronounced. The fibers in the SHG channel are collagen [172] while those in the TPF channel should be elastin [171]. Moreover, the globular structures shown in the TPF channel (green in Figure 5.8(d) and Supplementary Media 11) are speculated to be granular elastic tissues which are usually present in sun-damaged skin [203–205]. 91  The dorsal forearm of a 41-year-old Asian male volunteer with skin inflammation was also imaged under 730 nm excitation. The simultaneously acquired RCM and SHG + TPF images are shown in Figure 5.9(a) and Figure 5.9(b), respectively. False color overlay of SHG + TPF (green) and RCM (red), and false color overlay of SHG + TPF (red) and RCM (green) images can be found in Figure 5.9(c) and Figure 5.9(d), respectively. Comparing with normal skin shown in Figure 5.6, the size of the cells are larger and there are more space in between the cell boundary and cytoplasm, indicating the cells may be swollen because of the inflammation. This can be better seen in the false color overlaid image shown in Figure 5.9(d).  Figure 5.9: (a) RCM, (b) SHG+TPF, (c)false color overlay of SHG+TPF (green) and RCM (red), and (d) false color overlay of SHG+TPF (red) and RCM (green) images from the dorsal forearm of a 41-year-old Asian male volunteer. Excitation wavelength = 730 nm. FOV = 150μ m × 150μ m. Resolution = 256pixels×256pixels. Frame rate = 15 f ps. Scale bar: 20 μ m.  92  5.6 Discussion In this system, an avalanche photodiode (APD) module was used to collect the reflectance confocal signals. The reason is that compared to the photomultiplier tube (PMT) used in Chapter 3, the APD module has a much lower noise level, which can generate a much cleaner image. Another consideration is that the APD module is less sensitive to room light compared to the PMT detector used in Chapter 3, and therefore it allows for taking reflectance confocal microscopy imaging without turning the room light off, which makes the operation more convenient. In addition, since the previous frame grabber described in Chapter 4 drops frames and reduces the captured frame rate from 27 f ps to 24 f ps due to its latency, a new 10-bit multichannel frame grabber was used in this system design. The captured frame rate can now be as fast as 27 f ps. Moreover, because the new frame grabber can be operated in DC coupling mode, the image quality was also improved compared to using the old frame grabber which can only be operated in AC coupling mode. From the ex vivo SHG+TPF image shown in Figure 5.4(b), some of the cells show a bright circle inside the cell, which mimics the shape of a cell nuclear membrane. Moreover, some of the bright structures inside the cells are shown as disconnected features. The same structure is not shown under RCM imaging channel. Similar findings have not been reported in the literature. Interestingly, the cell membrane of each cell is not shown as bright under the SHG+TPF imaging modality. One possible explanation is that the nuclear membrane is a double lipid bilayer with perinuclear space in between, which is structurally different from cell membrane and may be able to generate higher TPF signals. Another possible explanation is that the bright features may represent special proteins only presented inside the nuclear membrane but not the cell membrane. In order to confirm the structure accurately, horizontal sectioning should be performed on excised skin samples. One skin section should be imaged using the developed MPM system under the same condition, and an adjacent section should be stained for histological analysis. By comparing images of the two adjacent skin sections, direct correlation can be made, which should explain the 93  structure contributed to the high SHG+TPF signals. Overall, RCM imaging channel shows a more limited depth of penetration and has a lower signal-to-noise ratio as compared to the SHG+TPF imaging channel when imaging skin dermis. Therefore, SHG+TPF imaging channel seems more suitable for imaging skin dermis. But RCM can still provide extra information. For example, cell boundaries are clearly seen in the RCM channel, while cytoplasm is better seen in the TPF imaging channel. At the dermis level, SHG and TPF channels show collagen bundles and elastin fibers, respectively. Other researchers have had success in reducing subject motion using a coverslip between the objective and the skin allowing image averaging over several seconds [178]. In this study, it was also confirmed that using a glass coverslip between the objective and the skin will help in reducing the motion artifacts, and allow us to average multiple frames for improving image quality. A plastic layer with a empty circular area in the middle was also compared with the glass coverslip setting (data not shown), and it was found that there were significantly higher amount of axial skin movement comparing with the coverslip scheme. The reason is that with a layer of glass coverslip on top of the skin, it provides extra pressure and prevents the skin from moving up and down in the Z axis. However, it is ultimately desirable to have imaging rates as fast as possible to satisfy the clinical requirements. As RCM and MPM are depth resolved techniques, to perform volumetric imaging, fast frame rates, such as the video frame rate demonstrated in this study, are essential.  5.7 Conclusion To summarize this chapter, the capability of simultaneous acquiring femtosecond reflectance confocal and multiphoton images of both ex vivo and in vivo human skin has been demonstrated. With the same scanning mechanism described in Chapter 4, up to 27 f ps frame rate can be achieved. By using the same laser source for RCM, SHG, and TPF imaging channels, we ensure that the images are well co-registered. This study also demonstrat-  94  ed that RCM and MPM images from the skin could provide complementary information. These unique RCM/MPM system features are important advances that will facilitate future clinical applications of RCM/MPM. It is envisioned that further development of RCM/MPM imaging will lead to instruments that can provide diagnostic information comparable to histopathology without the need for tissue excision.  95  Chapter 6 Raman Spectroscopy Combined with Reflectance Confocal Imaging and Multiphoton Microscopy Imaging System A version of Chapter 6 has been submitted for publication as: Wang, H., Lee, A.M.D., Lui, H., McLean, D.I., and Zeng, H. (2013), A method for accurate in vivo micro-Raman spectroscopic measurements under guidance of advanced microscopy imaging. Scientific Reports. Edits have been made to better meet the requirements of the journal.  6.1 Introduction Confocal Raman spectroscopy, which is a noninvasive optical technique, could provide micron-level resolution and depth-resolved biochemical information of in vivo biological tissues. This technique has been applied to study a number of in vivo health related phenomenon of human tissues, such as noninvasive assessment of human corneal hydration [206], estimation of stratum corneum thickness [207], and monitoring drug penetration  96  depth inside the skin [74]. Confocal Raman spectroscopy has also been proved to have potentials in helping early diagnosis of a variety of cancers [71, 133, 208, 209]. With the permissible laser power followed by ANSI standard [127], the usual acquisition time for a confocal Raman spectrum from biological tissue can be as long as 80 s [210]. During the tens of seconds of acquisition time of each spectrum, the acquired confocal Raman spectrum will likely contain information from out-of-target area due to movement of the subject. Reflectance confocal microscopy (RCM) imaging, which is capable of optical sectioning, has been considered to guide the spectral measurements. There have been some studies on combining reflectance confocal imaging with confocal Raman spectroscopy, but the imaging channel and the spectral acquisition are not from the same excitation laser source and the wavelengths are different [101]. Therefore, registration between the confocal image and the spot/area of Raman spectral measurement could be problematic. Moreover, most of the Raman spectral measurements were performed as single-point measurements, which require turning off the optical scanner to generate a stationary beam. Even co-registration of the sampling locations of the imaging and spectral channels are performed, with the movement from the subject during in vivo measurements, the confocal Raman spectrum will not necessarily come from the targeted point inside the tissue, but from an integrated area according to the random movement range from the subject [102]. In addition, RCM imaging is only able to provide morphological information based on refractive index variations while multiphoton microscopy (MPM), which employs two-photon fluorescence (TPF) and second-harmonic-generation (SHG) images, has proved to be able to provide complementary information on structures and biochemistry, and could be better utilized as a diagnostic tool [211]. Therefore, the hypothesis of this chapter includes (1) co-registered reflectance confocal imaging and confocal Raman spectral measurements using the same laser is achievable; (2) well-defined region-of-interest (ROI) confocal Raman spectral measurements can be performed without turning off the scanner and under the imaging guidance of reflectance  97  confocal microscopy; (3) integrating multiphoton microscopy could enhance the imaging capability. It is planned to validate this method by performing ROI confocal Raman spectral measurement under the guidance of reflectance confocal imaging and multiphoton microscopy on skin in vivo. It is expected that the results will stimulate great interests for other researchers to perform accurate, in vivo spectroscopic measurements under advanced imaging guidance. This study will also serve as a useful application of the proposed multimodality system.  6.2 System design and development A multimodal spectroscopic and imaging system has been developed as shown in Figure 6.2. Briefly, a 785 nm continuous-wave (CW) diode laser is used for both confocal Raman spectral measurement and CW reflectance confocal microscopy (cwRCM) imaging. The laser beam is first directed to an optical scanning system consisting of a resonance scanner and a galvanometer scanner, and then is focused by a 60X (NA = 1.0) water-immersion microscope objective onto human skin. A polarization beamsplitter (PBS) along with a quarter waveplate is used to direct the descanned reflectance confocal signals to an avalanche photodiode (APD) module with a 30 μ m pinhole in front. A dichroic beamsplitter is used to direct the Raman signals to a 50 μ m fiber connected with a Raman spectrometer. A femtosecond (fs) Ti:Sappire laser is used for multiphoton microscopic imaging. The fs laser beam is scanned by the same scanner and focused onto the skin by the same microscope objective. The TPF and SHG signals are collected by a photomultiplier tube to generate SHG+TPF images, and the descanned fs reflectance confocal microscopy (fsRCM) signals will be directed by another PBS onto an APD with a 50 μ m pinhole. Since almost all the optical materials that are transparent for visible light have a positive dispersion, which means that longer wavelengths travel faster through these materials. A double-prism based optical compressor, which could generate a negative dispersion by making the longer wavelength travel longer distance inside the second prism. Therefore,  98  Incoming pulses  D-shaped mirror Prism1  Compensated pulses  Prism2 Mirror  Figure 6.1: Diagram of optical prism compressor. to compensate for both dispersion inside the laser cavity and the dispersion generated by beamsplitters, microscope objective, and other optical components, a prism optical compressor, which consists of a pair of prisms and a mirror, was built in the excitation arm (shown in Figure 6.1). A D-shaped mirror was used to direct the compensated pulses to the MPM system. A flip mirror is employed to switch between the Raman excitation CW laser and the MPM excitation fs laser, and the two beam paths have been optimized to be the same. To reduce involuntary body movement, double sided tape and a metal ring were used to magnetically mate the skin with the imaging objective. Water was used in between the microscope objective and the skin surface. The Raman laser power incident on the skin was 27 mw, and the fs laser power incident on the skin was adjusted to be ≤ 40 mW using a half-waveplate/polarizer combination at the laser exit. The imaging field of view (FOV) can vary from 10μ m ×10μ m to 300μ m ×300μ m. The study was approved by the University of British Columbia Research Ethics Board (#H96−70499). Informed consent was obtained from the volunteer subject.  6.3 Measurements on in vivo human skin Since an optical compressor was introduced into the system, higher efficiency of generating the multiphoton signals is expected. A combined TPF and SHG video (Supplementary  99  M  M  M  M  l/4 WP Laser block  Prism2 fs laser  L  L  M  L Scanner Dichroic  l/2 WP  Flip Mirror  Prism1 L  PBS  LP  Microscope Objective M  L  SP  SHG+TPF PMT  Dichroic  PBS  M PBS052  D-shape Mirror L  M  M l/2 WP  M L  BP  M  L Human skin  L L  Pinhole  LP  785nm CW laser  Pinhole  Raman Spectometer Fiber holder Optic fiber  fsRCM APD  cwRCM APD  Figure 6.2: System diagram of in vivo multimodal confocal Raman spectroscopy, reflectance confocal and multiphoton imaging. WP: waveplate; PBS: polarization beamsplitter; M: mirror; APD: avalanche photodiode; L: lens; LP: long-pass filter; SP: short-pass filter; PMT: photonmultiplier tube. Media 12) on in vivo volunteer forearm skin was acquired to characterize the improvement. Three representative image frames from the video are shown in Figure 6.3. Clear granular shaped keratinocytes can be visualized in Figure 6.3(a) and more details of the cellular structures can be seen as compared to Figure 4.4 in Chapter 4. Bright basale cells and dermal fibers can also be visualized as shown in Figure 6.3(b) and (c). The validation experiment performed in this study was to measure different microstructures of a cherry angioma lesion in vivo on the upper arm of an Asian male volunteer. A dermoscopic image was first taken with DinoLite (AD4013T−TVW, AnMo Electronics Corporation, Hsinchu, Taiwan) to record the clinical appearance of the lesion as shown in Figure 6.4(a). Both fsRCM and SHG+TPF imaging channels were used to locate a blood vessel inside the cherry anigoma lesion as the 1 st example ROI with an imaging FOV of 300μ m×300μ m. The flip mirror was then flipped on to allow cwRCM imaging and confocal Raman spectral acquisition. The 300μ m×300μ m FOV cwRCM image is shown in Figure 6.4(b). Then, the FOV of the cwRCM imaging channel was shrunk to 100  (a)  (b )  (c)  Figure 6.3: In vivo integrated SHG+TPF images extracted from Supplementary Media 12 of the dorsal forearm of a 23-year-old Asian male (optic compressor introduced into the system).(a) stratum granulosum, (b) stratum basale, and (c) dermal fibers. Excitation wavelength =750 nm. FOV = 150μ m×150μ m. Resolution = 256pixels×256pixels. Frame rate = 15 f ps. only cover the ROI shown in Figure 6.4(c), which is the same area shown as the red square in Figure 6.4(b). A confocal Raman spectrum was then taken under the guidance of the cwRCM imaging. A demonstration video of a process of finalizing the ROI for spectral measurement can be found in Supplementary Media 13 (a more complicated case than of Figure 6.4). The acquisition time for one spectrum is 20 s. The acquired confocal Raman spectrum is shown in Figure 6.4(d). The accompanied cwRCM images were also recorded as a video during the confocal Raman spectral measurement which can be found in Supplementary Media 14. The movement from the volunteer can be seen, but clear blood structures are shown in the video (Supplementary Media 14) during the whole acquisition period, indicating that the confocal Raman spectrum is valid from an ROI completely and always within the blood vessel itself. Strong peaks from hemoglobin (752 cm −1 ), glucose (1123 cm−1 and 1343 cm−1 ), and protein (940 cm−1 and 1665 cm−1 ) can be found on the Raman spectrum. As examples, we have rejected Raman measurement results where the videos show that there are moments when the ROI had moved outside of the blood vessel. To validate the method further, surrounding fiber structures and cellular structures were selected as the 2nd and 3rd example ROI, respectively. Figure 6.5(a) and Figure 6.5(e) were first taken at a FOV of 300μ m×300μ m from the SHG+TPF and fsRCM imaging  101  Figure 6.4: In vivo confocal Raman spectra of blood in a cherry angioma acquired under guidance of reflectance confocal imaging and multiphoton imaging. (a) Dermoscope image of the cherry angioma lesion; (b) cwRCM image of the blood vessel, image field of view (FOV) is 300μ m×300μ m; (c) cwRCM image extracted from a video (Supplementary Media 14) of the blood vessel of the region of interest (ROI) from the area of the red square in (b), the image FOV is 100μ m×100μ m; (d) confocal Raman spectrum of the blood vessel of the area shown in (c) and (b) as red squares, the exposure time is 20 s. channels, respectively. The cwRCM images shown in Figure 6.5(b) and Figure 6.5(f) were 150μ m×150μ m FOV and were from the green squares labeled in Figure 6.5(a) and Figure 6.5(e). The confocal Raman spectra of the fiber structures and cells (shown in Figure 6.5(d) and Figure 6.5(h), respectively) were acquired from the region with a FOV of 60μ m×60μ m shown in Figure 6.5(c) and Figure 6.5(g) as red squares. Strong Raman peaks at 856 cm−1 , 937 cm−1 , and 1454 cm−1 representing collagen and elastin are shown in the confocal  102  Raman spectrum in Figure 6.5(d). Raman peaks of cell nuclei (722 cm−1 and band 13251330 cm−1 ), protein (1450 cm−1 ), phenylalanine (1003 cm−1 ), and lipids (1655 cm−1 ) can be found in the confocal Raman spectrum shown in Figure 6.5(h). Raman peak assignments are listed in Table 6.1. Table 6.1: Raman peak assignment for in vivo human skin with a cherry angioma lesion. Peak position (cm−1 ) Assignment 722 DNA 752 Hemoglobin 787 Nucleic acids 856 Collagen 940 Collagen, proline, hydroxyproline 1003 Phenylalanine 1124 Glucose 1325-1330 Nucleic acids 1343 Glucose 1450 Protein and lipids 1454 Elastin 1558 Tryptophan 1618 Tryptophan 1665 Amide I  Reference [212] [213] [214] [138] [138] [133] [137] [135] [136] [68] [215] [138] [141] [138]  The fsRCM and cwRCM images clearly show blood flow inside each vessel, and single blood cells can be identified. It is noticed that the blood flow velocity inside the cherry angioma lesion varies significantly from vessel to vessel. Therefore, the high quality cwRCM images from the blood vessel allow us to derive the blood flow velocity by simply tracking a single blood cell. For example, Figure 6.6 shows 4 consecutive frames from a blood flow video (Supplementary Media 13, frame19-22). The target blood cell labeled with a yellow arrow in Figure 6.6 (frame f20) was not observed in Figure 6.6 (frame f19), but its trajectory can be seen in the next two frames (f20 and f21) shown in Figure 6.6 (frame f21) and Figure 6.6 (frame f22). The distance that the cell moved from f21 to f22 was 0.03mm, and the time difference between each frame is 1/15 s. Therefore, the velocity can be calculated  103  Figure 6.5: In vivo confocal Raman spectra of cells and fibers surrounding a cherry angioma acquired under guidance of reflectance confocal imaging and multiphoton imaging. (a) SHG+TPF image of the fiber structures surrounding a blood vessel of the cherry angioma lesion, image field of view (FOV) is 300μ m×300μ m; (b) Zoom-in cwRCM image of the area shown in the green square of (a), image FOV is 150μ m×150μ m; (c) cwRCM image showing the region of interest (ROI) from the area of the red square in (b), the image FOV is 60μ m×60μ m; (d) confocal Raman spectrum of the fiber structures of the red square area shown in (c) and (b), the exposure time is 20 s; (e) fsRCM image of the cellular structures surrounding a blood vessel of the cherry angioma lesion, image FOV is 300μ m×300μ m; (f) Zoom-in cwRCM image of the area shown in the green square of (e), image FOV is 150μ m×150μ m; (g) cwRCM image showing the ROI from the area of the red square in (f), the image FOV is 60μ m×60μ m; (h) confocal Raman spectrum of the cellular structures of the area shown in (f) and (g), the exposure time is 20 s. out as 0.45 mm/s. Interestingly, the glucose peaks (1124 cm−1 and 1343 cm−1 ) can be clearly found in the confocal Raman spectra acquired at the blood vessel (Figure 6.4(d)). To explore whether changes of blood glucose level can be noninvasively detected, an in vivo spectrum from the blood vessel of the cherry angioma lesion was first taken with an integration time of 20 s as zero-point reference. The volunteer was then asked to drink a standard glucose liquid (75 g glucose tolerance test beverage, Thermo Fisher Scientific Inc., Waltham, MA) within 5 min. Confocal Raman spectra were taken every 15 min for the next 105min time. The highest glucose level was found between 45 to 60 min after the volunteer took the glucose  104  f19  f21  f20  f22  Figure 6.6: Four consecutive video frames used to derive blood velocity (Supplementary Media 13). Yellow arrow points to the targeted blood cell in f20, f21, and f22. Image FOV is 300μ m×300μ m. liquid, with ∼ 10% increase compared with zero-point reference. The change of glucose level is plotted over time by taking ratio of the area under the glucose peak at 1124 cm −1 and the area under the protein peak at 1450 cm−1 , which is considered as a stable Raman signal in the blood, not affected by glucose concentration (Figure 6.7).  6.4 Discussion It is believed that the method demonstrated in this chapter is a powerful tool to perform accurate region-of-interest confocal Raman spectral measurements under the guidance of reflectance confocal and multiphoton microscopic imaging. This is especially critical 105  Figure 6.7: In vivo monitoring of changes of blood glucose level of a cherry angioma lesion on the upper arm skin of a volunteer using confocal Raman spectroscopy guided with reflectance confocal imaging. when performing measurements on in vivo subjects because the capability to evaluate the movement level during the measurement is a key factor for accurate interpretation of the acquired spectral data. We did find that sometimes there were severe movements during the measurement, the spectrum then contained contributions from the outside of the ROI. In this case, we will simply re-take the measurement one more time to make sure the level of movement from the subject is small enough to be negligible. However, if there is no real-time monitoring from an imaging modality, the qualitative or quantitative analysis based on the spectral data will result in inaccurate conclusions. For example, when using confocal Raman spectroscopy to study skin cancer, a spectral measurement which should target on cancer cells only, may contain both normal and cancer cells due to the movement from patients, and the only way to exclude the invalid spectral data is the real time imaging guidance. Therefore, imaging guidance is crucial for in vivo biological micro-Raman spectral measurements. The flexibility of changing the area of ROI from 300μ m×300μ m to 10μ m×10μ m can  106  easily allow us targeting on structures of different sizes. With the help from multiphoton microscopic imaging, both morphological and biochemical information of biological tissues can be obtained. For the specific experiment on the cherry angioma measurements, SHG and TPF signals were not separated because the focus of this study was not to target on separating collagen from elastic fibers. But with simple instrument modification as already shown in Chapters 4 and 5, SHG and TPF signals can be well separated and be used for measuring specific fiber structures in the dermis. It was also found that reflectance confocal imaging is proper for imaging blood flow in vivo. When deriving the blood flow velocity, the factor that the blood cell may not exactly move within the screen plane was no considered. The reason is that the travel distance between the 2 consecutive frames was relatively short, and the difference of the blood flow direction therefore may not be crucial. With further improvement on imaging penetration depth, the capability of deriving blood flow may be applicable to noninvasively analysis of blood related diseases. It is very interesting to find clear glucose peaks in the confocal Raman spectra of blood with greatly reduced interferences from other tissue components. Although the idea of using Raman spectroscopy to noninvasively derive glucose level is not new [216], these measurements may provide a few novel and valuable insights. First, under confocal imaging guidance, it is ensured that the glucose information is only from the blood vessel, not elsewhere in the tissue. This improves the accuracy of noninvasively measure the blood glucose level. Secondly, the Raman spectral quality of our measurement is significantly improved compared with results reported in the previous literature [216, 217]. This is probably due to our confocal measurement geometry, precise blood vessel targeting under imaging guidance, and the high sensitivity of our spectrometer. Thirdly, other important micro-scale localized biochemical information such as the level of oxyhemoglobin and deoxyhemoglobin may also be derived using the method described in this study. In the experiment performed in Chapter 4 and Chapter 5, the capability of differenti-  107  ating collagen from elastic fibers in the skin using SHG and TPF imaging modalities has been demonstrated. Therefore, the SHG and TPF imaging channels were not separated in this study. Rather, the focus of this chapter is to demonstrate the method of real-time monitoring for spectral measurements, and to illustrate the advantageous of having all the modalities integrated in one system.  6.5 Conclusion In conclusion, the described method of real time monitoring in vivo confocal Raman spectral measurements with reflectance confocal and multiphoton imaging could provide more accurate spectral interpretations. The proposed method has been validated by measuring different structures in vivo on a volunteer. The demonstrated ROI method is valuable for precise in vivo spectral measurements and the imaging modalities allow noninvasively obtaining information on the blood flow velocity. In addition, the special non-invasive method of obtaining blood glucose level is novel and practical to be adapted into clinical use. Therefore, it is believed that the developed method can be generally applied into various types of in vivo micro-spectroscopy measurements. One should perform micro-spectroscopy measurements carefully under advanced microscopy imaging guidance in order to generate accurate spectral interpretation of the biological, physiological, and biomedical phenomena under investigation.  108  Chapter 7 Imaging Directed Photothermolysis through Two Photon Absorption A version of Chapter 7 has been published online as: Wang, H., Zandi, S., Lee, A.M.D., Zhao, J., Lui, H., McLean, D.I., and Zeng, H. (2012), Imaging directed photothermolysis through two-photon absorption demonstrated on mouse skin − a potential novel tool for highly targeted skin treatment. Journal of Biophotonics, DOI: 10.1002/jbio.201300016. Edits will be made to better meet the requirements of the journal.  7.1 Introduction Lasers can be used to target skin chromophores such as water, hemoglobin, and melanin to perform fast and precise treatment on a variety of skin lesions [218, 219]. Laser therapy generates little discomfort during and after the treatment, and also has lower risk of scarring. Common skin conditions that can be treated using laser therapy include: vascular lesions (port-wine stains, haemangiomas), pigmented lesions (freckles, birthmarks, nevi), tattoos, unwanted facial or body hair, facial wrinkles, sun-damaged skin, keloids, hypertrophic scars, and skin cancers [220–222]. Traditional laser treatment, which is essentially a one-photon absorption process, has  109  chances of causing unintended tissue damage at microscopic level. For example, if the targeted chromophore is the melanin inside dermis, by focusing the laser beam onto the targeted melanin, the untargeted melanin inside epidermis will highly likely to be damaged as well due to the absorption to the laser light. This scenario is relevant for laser treatment of dermal tumors, hirsutisum, hyperhidrosis, etc. Two-photon absorption (TPA) is a process where two near infrared (NIR) photons get absorbed simultaneously by a molecule, and it only occurs at the focal point. This provides the possibility of highly targeted tissue alteration of specific dermal structures without damaging structures in the epidermis, and may potentially solve the problem in the one-photon laser treatment. Interestingly, during the TPA process, there are natural fluorophores inside the skin, such as keratin, NADH, melanin, and elastin, which can emit two-photon excitation fluorescence (TPF), while non-centrosymmetric structures such as collagen, can produce secondharmonic-generation (SHG) signals [197]. This leads to high biochemical specificity detection. Therefore, TPF and SHG signals can be used to image before and after the twophoton laser-induced tissue alteration, and also for monitoring tissue alteration during the high power irradiation. More advantages of two-photon excitation include its inherent optical sectioning capability, deeper penetration depth, less photo-damage and photo-bleaching of the non-focused areas on the beam path. The hypothesis of this chapter is that TPA from the skin will provide precise damage to targeted skin structures by inducing two-photon absorption photothermolysis. The exploration of two-photon absorption based laser therapy has been growing recently. Two-photon photodynamic therapy has been demonstrated using NIR femtosecond (fs) pulses [223]. It is a photochemical process which involves using two-photon absorbing dye as a photosensitizer, and is mainly used in cancer treatment. To evaluate two-photon based photo-thermal effects, an YB: KYW femtosecond (fs) laser has been used to ablate porcine corneal samples. Ablation thresholds were also determined using three diode pumped solid-state ultrafast lasers. It was found that corneal ablation threshold remained  110  almost constant within the first 200 μ m of stroma and was proportional to the square root of the laser pulse width [224]. The relationships between the fs laser pulse length, power density, pulse number, and the cornea excision quality as well as the resulting tissue morphologies have also been studied using visible laser wavelengths [225]. However, to our knowledge, there has been no published study on TPA based laser treatment on skin. Therefore, the objective of this chapter is to explore the feasibility of two-photon laser absorption for targeted skin modification using the developed multimodality instrument. It is believed that two-photon laser absorption for targeted skin alteration may potentially be used to precisely treat dermal tumors, hirsutism, hyperhidrosis, and birthmarks without damaging the surrounding normal skin or causing scars. This study is also served as an example to demonstrate potential applications of the developed multimodality system.  7.2 Materials and methods 7.2.1 Animal preparation Skin excised from the shaved backs of euthanized female C3H/HeN mice (N=10) was used in this study. The skin was oriented on a glass slide with the epidermal side up and then covered with a glass coverslip. Distilled water was added in between the coverslip and the microscope objective for refractive index matching. All animal experiments were performed according to a protocol approved by the University of British Columbia Committee on Animal Care (certificate #: A10 − 0338).  7.2.2 Integrated reflectance confocal microscopy and two-photon imaging system The details of the home-made integrated reflectance confocal (RCM) and multiphoton microscopy (MPM) system was used for the laser treatment experiment and for imaging guidance and monitoring. Details of the system can be found in Chapter 5 and in our published paper [211]. Briefly a tunable, fs Ti: Sapphire laser (720-950 nm, 140 f s pulsewidth, 90  111  MHz, Chameleon, Coherent Inc., Santa Clara, California) was used as the excitation source for both the tissue alteration and the imaging monitoring. To improve both the illumination and detection efficiency, the 50/50 beamsplitter used in Chapter 5 was replaced with a polarization beamsplitter (PBS) along with a quarter waveplate to effectively direct both the illumination laser beam to the sample and the de-scanned reflectance confocal signals to the RCM detector. The illumination laser beam was circular-polarized and directed into a 60X (N.A. = 1.0) water-immersion microscope objective and focused onto the skin. The laser spot at the skin was scanned using two scanners to generate the RCM and SHG+TPF images with a speed of 0.067 second per frame (15 frames per second). Both TPF and SHG signals were collected onto a single photomultiplier tube (PMT), hereinafter referred to as the SHG+TPF imaging channel. The RCM and SHG+TPF images were recorded by a frame grabber as synchronized independent video streams. As both the RCM and SHG+TPF imaging channels share the same laser, scanners, and microscope objective, the RCM and SHG+TPF images are perfectly registered [211]. A polarizer-λ /2 waveplate combination at the laser exit was used to adjust the incident laser power at the sample to be 30 mW for imaging, and 75 mW or 200 mW for tissue alteration/treatment. The wavelength used for both imaging and tissue alteration was 785 nm. This wavelength was selected because: (1) it falls within the tissue optical ”window”, where the one-photon absorption of light by tissue is minimized and tissue penetration depth is maximized; (2) it matches the wavelength of our existing CW laser that will be used in the CW and fs laser comparison experiment. TPA at 785 nm corresponds to the one-photon absorption in the UV wavelength range, which has been employed in conventional one-photon phototherapy due to strong tissue absorption. In order to target at different depths, the microscope objective was mounted on a piezoelectric scanner (MIPOS 500 SG, Piezosystem Jena, Jena, Germany) that allows up to 400  μ m of closed-loop travel. Coarse adjustment of the scanning area was achieved using a manually actuated 3-axis translational stage.  112  Both RCM and SHG+TPF images were acquired before and after the tissue alteration at the targeted skin layers as well as at vertical levels of 0 μ m and 60 μ m depth within the skin. For all the tissue alteration processes, 0 μ m depth was defined as the layer where the uppermost stratum corneum just can be found in the RCM imaging channel. During the high power treatment irradiation, RCM and SHG+TPF were also used to monitor the damaging process. In total, 20 pieces of mouse skin were imaged and treated. Three different target depths beneath the skin surface (20, 30, and 40 μ m) and three different exposure times (from 0.5 to 3 min) were tested to study the effects of target depth and irradiation period on tissue treatment efficacy (Table 7.1). Figure 7.1 illustrates the horizontal views of the skin sample (X-Y plane is parallel to the skin surface, Z represents the depth into the skin). The skin sample blocks are 5mm ×10 mm size. In order to orient and localize the target damaged sites for subsequent histologic analysis the irradiation processes were initiated over areas of 150μ m × 150μ m at 1 mm away from one of the long edges of each sample (Figure 7.1(b)). Then, following the irradiation parameters for each specimen group, the skin was moved 150μ m along the Y-axis in order to treat the adjacent area (150μ m × 150μ m) at the same depth Z. By iterating this process, a line of damage along the Y-direction could be created within an X-Y plane at certain depth in the dermis that was 1 mm away from both edges. After each irradiation process, the epidermis was marked twice in parallel to each side of the laser-induced damaging line using a scalpel (Figure 7.1(a)). The skin samples were then processed, sectioned, and stained with hematoxylin and eosin (H&E) for histologic examination (Figure 7.1(c)). The damage line induced by the high power laser irradiation, the cutting mark lines, and the histological sectioning are schematically illustrated in Figure 7.1. Figure 7.1(a) demonstrates the two epidermal cuts, and Figure 7.1(b) shows the line of damage in the dermis with each exposure unit area of 150μ m × 150μ m. The skin samples were sectioned perpendicularly to the epidermal cuts and line of tissue damage. This allowed us to locate the damaged area in the dermis with two epidermal cuts as reference in the histological slides.  113  Cut2  Cut1 Epidermis  (a)  5  Dermis  mm  1  mm  1  mm  (b)  Line of tissue damage  Sectioning  (c)  Y 10  X  mm  Figure 7.1: Diagram demonstrating the horizontal view (X-Y plane) of the epidermal cuts, the line of tissue damages at dermis layer, and the histological sectioning. The three skin-colored rectangular blocks represent the horizontal view of the skin samples, which were 5 × 10 mm size. (a) Horizontal view of epidermal layer of the skin with two cuts; (b) Horizontal view of dermal layer with the line of tissue damage, which consists of several exposure areas shown as red squared blocks; (c) Horizontal view of sectioning direction on the X-Y plane. Multiple sections were performed in order to cover the whole line of damage. In order to test whether the tissue damages generated by the ultrafast fs laser are mainly due to two-photon absorption versus single photon near infrared absorption, a control study using a continuous wave (CW) laser was performed. The CW laser (I0785SA0100B − T K, Innovative Photonics Solutions, Monmouth Junction, New Jersey), which has a central  114  Table 7.1: Combinations of target depth and laser exposure duration. Skin sample # Target depth (μ m) 1-3 20 4-5 30 6-7 30 8-9 30 10 30 11-13 40 14-16 40 17-18 40 19-20 40  Irradiation period (min) 0.5 0.5 1 2 3 0.5 1 2 3  wavelength of 785 nm, was coupled into the combined reflectance confocal microscopy and two-photon imaging system using a flip mirror. This set-up allows us to switch between the CW laser source and the fs laser source. The CW laser beam path was optimized to follow the same fs laser beam path, allowing the two laser beams to focus at the same depth within the skin. In this way the reflectance confocal imaging derived from the CW laser was used to monitor the CW laser-induced tissue alteration itself, and SHG+TPF imaging from the fs laser was used before and after the CW irradiation to better evaluate the tissue damages. For the CW laser, the maximum power that can be coupled into the microscope objective was 60 mW . Therefore, in the control experiments, for both the CW and fs laser-induced tissue alterations, the following parameters were the same: 60 mW power, 785 nm wavelength, 30 μ m tissue treatment depth, and 70μ m × 70μ m exposure area. For the fs laser-induced tissue treatment experiment, confocal and SHG+TPF images of the skin were acquired before, during and after each high-power irradiation, whereas for the CW laser-induced tissue treatment experiment only CW confocal images were acquired during the tissue treatment process, since there is no SHG and TPF signals generated in the CW mode. Nevertheless paired SHG+TPF images generated from the fs laser and CW confocal images were acquired before and after each high-power irradiation process. This comparison experiment was performed at multiple sites on 3 skin samples.  115  7.3 Results SHG+TPF  RCM  RCM_Before  RCM_After  SHG+TPF_Before  SHG+TPF_After  0s Above target (0mm) 1  3s Target area (20mm)  6s Below Target (50mm)  9s  (b) (a)  Figure 7.2: RCM and SHG+TPF images comparing before, during and after high power fs laser irradiation. Laser wavelength: 785 nm, imaging power: 20 mW , irradiation power for tissue treatment: 75 mW , target depth: 20 μ m, irradiation period: 0.5 min, both the treatment area and image field-of-view (FOV) are 70μ m × 70μ m. (a) RCM and SHG+TPF image frames at 0, 3, 6, and 9 s during the laser irradiation; (b) RCM and SHG+TPF image frames extracted from the tissue damaging video (Supplementary Media 15 and 16) at 0, 20, and 50 μ m depths. In order to directly monitor the tissue effects, both RCM and SHG+TPF images were acquired during the high power laser irradiation treatment. In the dermis, the main fluorophore that generates TPF is elastin, and collagen fiber is the structure that generates strong SHG signals. Figure 7.2(a) shows a sequence of extracted RCM and SHG+TPF video images at 4 different time points during treatment where the targeted depth was 20 μ m (Supplementary Media 15 for RCM channel, and Supplementary Media 16 for SHG+TPF channel). The dermis was intact in both RCM and SHG+TPF imaging channels at 0 s baseline. With 75 mW of power irradiation, dermal fiber alteration first became apparent at 3 s. By 9 s obvious changes were seen in both the RCM and SHG+TPF imaging channels. It is believed that during the laser treatment, both elastic and collagen fibers were 116  denatured due to the increased temperature, which accounted for the loss of SHG+TPF signals [226, 227]. Figure 7.2(b) shows representative RCM and SHG+TPF images comparing before and after targeted treatment as monitored at different tissue depths. Above the targeted layer, the stratum corneum remains qualitatively the same in the RCM imaging channel at the completion of the laser exposure. Below the 20 μ m target layer (i.e. at ∼ 50 μ m beneath the 0 μ m surface reference layer), both RCM and SHG+TPF imaging channels showed strong signals from dermal fibers that were morphologically very similar before and after treatment. At the targeted layer (∼ 20 μ m beneath the 0 μ m reference layer), fibrous structures were found in both RCM and SHG+TPF images before treatment. During the laser treatment, it was found that in contrast to the SHG+TPF imaging channel which shows dark holes within the target layer, the RCM signals were increased in the form of bright structures and focal signal saturations. RCM and SHG+TPF at the level of the targeted skin layer appear different as monitored throughout the laser exposure. The marked epidermal landmarks (two cuts indicated by black arrows) were found on the hematoxylin and eosin (H&E) stained section shown in Figure 7.3(a). An empty space devoid of dermal fibers measuring approximately 100μ m × 50μ m (inside the blue circle) can be visualized between the two epidermal landmarks and represents the laser target site within the skin. Zoomed-in views of the two landmarks as well as the targeted area are shown in Figure 7.3(b), (c), and (d), respectively. Around the vicinity where the laser created an empty space, the adjacent epidermis, deeper dermis and subcutaneous fat layers remained intact. Figure 7.4 shows SHG+TPF images at different depths comparing before and after the high power (200 mW ) laser irradiation with a target depth of 40 μ m and a tissue treatment time period of 2 min. No significant structural changes were found at 10, 20, and 30 μ m depths. At 40, 50, and 60 μ m depths, an empty hole in between the fiber structure was seen. To explore the correlation between the irradiation period and tissue treatment efficacy,  117  500 m m  (a)  100 m m  (b)  (c)  (d)  Figure 7.3: H&E histological photomicrographs demonstrating selective alteration in laser-exposed mouse skin: (a) Histology view of the damaged area (indicated by the blue circle) and the two epidermal cuts (indicated by the black arrows). Zoomed-in views of (b) left cut (indicated by the black arrow), (c) targeted damaged area (indicated by the blue circle), and (d) right cut (indicated by the black arrow). SHG+TPF images at different depths comparing before and after the procedure with a target depth of 40 μ m and irradiation periods of 1 min and 3 min were also acquired (data not shown). For 1 min irradiation period at 40 μ m target depth, only very subtle changes were found in the images. However, for 3 min irradiation period at 40 μ m target depth, much bigger holes in between the dermal fibers were observed at 40, 50, and 60 μ m depths with no visible changes at the upper layers. To determine whether targeting at different depth will have different effects at other depths, high power irradiation targeting at 30 μ m beneath the skin surface was also performed. Figure 7.5 shows SHG+TPF images of different depths comparing before and  118  after  before 0mm  0mm  10mm  10mm  20mm  20mm  30mm  30mm  40mm  40mm  50mm  50mm  60mm  60mm  Figure 7.4: SHG+TPF images of different depths comparing before and after high power (200 mW ) laser irradiation with a target treatment depth of 40 μ m and a irradiation period of 2 min. The exposure area and the image FOV: 150μ m × 150μ m.  119  before  after 0mm  0mm  10mm  10mm  20mm  20mm  30mm  30mm  40mm  40mm  50mm  50mm  60mm  60mm  Figure 7.5: SHG+TPF images of different depths comparing before and after high power (200 mW ) irradiation with a target depth of 30 μ m and an irradiation period of 2 min. Exposure area and the image FOV: 150μ m × 150μ m.  120  after the high power (200 mW ) laser irradiation with a target of 30 μ m and with an irradiation period of 2 min. Significant structural dermal changes were found starting from 20 μ m. Multiple holes in between the dermal fibers were observed at depth of 30, 40, 50, and 60 μ m. Layers above 20 μ m depth were left intact. SHG+TPF images of different depths comparing before and after the procedure targeting 30 μ m beneath the surface with an irradiation period of 0.5 min were also acquired (data not shown). Similar effects were observed compared to the results achieved with an irradiation period of 2 min. Dermal changes were found from 20 μ m down to 60 μ m. Epidermal changes were not observed. In order to confirm that the tissue alterations generated by fs laser pulses are dominated by two-photon absorption rather than one-photon absorption, a control study using a CW laser was performed. The CW laser irradiation will induce one-photon absorption only. The target depth was selected as 30 μ m. Before the high power irradiation (60 mW ), a CW confocal image was taken which is shown in Figure 7.5(a-1). Since MPM images are better for demonstrating tissue damage, a SHG+TPF image was also taken before the procedure, which is shown in Figure 7.5(a-3). The high power irradiation was performed using the CW laser at 60 mW for an exposure period of 1.5 min, FOV (70μ m × 70μ m). No change in the CW confocal images was found during the irradiation process. As shown in Figure 7.5(a-2) and (a-4), intact fiber structures can be seen in both the CW confocal and SHG+TPF images after the procedure. However, with the same target depth (30 μ m), wavelength (785 nm), power level (60 mW ), FOV (70μ m × 70μ m), and irradiation period (1.5 min), significant dermal fiber damages were found after the fs irradiation. SHG+TPF images at different depths before the irradiation can be found in Figure 7.5(b) column (b-1), and the SHG+TPF images after the irradiation are shown in Figure 7.5(b) column (b-2). At the target depth of 30 μ m, damaged fibers can be seen clearly in the SHG+TPF images. No damage was found at the epidermis (0 μ m) and depth of 60 μ m. The fs confocal images before and after high power irradiation are shown in Figure 6 (b) column (b-3) and (b-4). Changes at depth of 20, 30, and 40 μ m can be found in the confocal images, but with  121  intact epidermis (0μ m) and intact dermal structures at depth of 60 μ m. This experiment comparing the fs laser treatment effect with that of the CW laser treatment demonstrated that at 785 nm, the one-photon absorption is weak and has minimum effect on tissue, while the fs laser treatment effect on tissue is indeed caused by strong two-photon absorption. fs laser irradiation treatment RCM  MPM Before  After  Before  After  0mm  CW laser irradiation treatment (a-1)  20 m m  After  Before (a-2)  30mm depth, RCM  30 m m (a-3)  (a-4)  30mm depth, MPM  40 m m  (a)  60 m m  (b-1)  (b-2)  (b-3)  (b-4)  (b)  Figure 7.6: Comparison of CW laser-induced tissue alteration (a) and fs laser-induced tissue alteration (b). CW confocal images of mouse skin at the target depth of 30 μ m showing before high power (60 mW ) irradiation (a-1) and after (a-2); SHG+TPF images generated from fs laser at the target depth of 30 μ m showing before high power irradiation (a-3) and after (a-4); SHG+TPF images of mouse skin at different depths showing before high power (60 mW ) irradiation in column (b-1) and after in column (b-2); fs confocal images of mouse skin at different depths showing before high power irradiation in column (b-3) and after in column (b-4). Exposure area and image FOV: 70μ m × 70μ m.  122  7.4 Discussion The experiment has clearly demonstrated the advantages of using two-photon absorption based photothermolysis on mouse skin treatment, including highly-targeted tissue alteration and the capability of real-time monitoring through RCM and SHG+TPF. For example, the results have shown that targeted damages of mouse dermal fibers without affecting epidermal structures can be achieved. RCM and SHG+TPF imaging channels can provide both morphological and biochemical information of the tissue structure under irradiation, which allow us to monitor the tissue alteration in real time. This feature can also be used to explore optimal tissue alteration efficacy. Moreover, from a clinical treatment point of view, timely stop of the irradiation procedure could be applied with these real-time imaging monitoring. This may potentially prevent incorrect treatment or overtreatment to the skin, therefore making the treatment process more controlled and flexible. Therefore, further improvement on the system performance such as improving the signal-to-noise ratio should be considered. It was noticed that some mouse skin samples, such as the one shown in Figure 7.2, have very weak signals in the SHG+TPF imaging channel, probably due to the thin stratum corneum and epidermis. Only one layer of cells was found in the histology images, and this could be the reason that the SHG+TPF imaging channel was lack of signals as shown in Figure 7.2. Interestingly, in Figure 7.4, the biggest diameter of the damaged area was found to be at the depth of 60 μ m, rather than the target depth at 40 μ m. A possible explanation was proposed as shown in Figure 7.7. Compared with before high power laser treatment, the skin could elevate due to the space generated during the procedure. In this case, if 0 μ m is still defined as the layer of stratum corneum, the largest diameter of the hole would be at a depth deeper than 40 μ m. From our experiments, it seems that for the target depth of 40 μ m, the tissue elevation was around 20 μ m, therefore, the largest diameter of the hole was at 60 μ m beneath the surface. For the target depth of 30 μ m, the tissue elevation was around 10 μ m, and the largest diameter of the hole was located at around the depth of 40 123  0mm  0mm  20mm  20mm  40mm  40mm  6 0mm  60mm  80mm after irradiation  before irradiation  Figure 7.7: Proposed explanation for tissue elevation during tissue alteration.  μ m, which can be found in Figure 7.5. When comparing the tissue alteration results for different target depths, it was noticed that the damage threshold for 40 μ m was much higher than 20 μ m or 30 μ m. To be more specific, for target depth at 40 μ m, dermal damages were observed starting from 20 s. However, for target depth at 30 μ m, dermal damages were found within 10 s of laser irradiation. For target depth at 20 μ m, dermal damages were initialized within 3 s of laser irradiation. In addition, for target depth at 30 μ m, it seems that the damaging is more likely to affect the skin layer at 10 μ m above the targeted layer. No discernable change was found at 10 μ m above the targeted layer when the target depth was set to be 40 μ m. This may indicate that photothermal effect at deeper depths is more restricted to local area compared with more superficial skin layers due to the structural differences at different depths. Therefore, this two-photon absorption based photothermolysis may be applied better to treat skin lesions at deeper locations. If the tissue alteration induced by fs laser was predominantly due to NIR one-photon absorption, similar tissue alteration effects with a CW laser should be observed. Therefore, targeted at the same target depth (30 μ m) and with the same wavelength (785 nm), power level (60 mw), FOV (70μ m × 70μ m), and irradiation period (1.5 min), CW laser-induced tissue alteration which represents one-photon absorption situation, and fs laser-induced  124  tissue alteration which represents two-photon absorption situation, were compared. The results showed that the CW laser failed to generate damaging effects, while significant dermal fiber damages were found during the fs laser irradiation under the same condition. This confirmed that the fs laser-induced highly-targeted tissue alteration that was observed in this study is not due to near infrared (NIR) one-photon absorption, but due to two-photon absorption as expected. And the two-photon absorption cross-section must be much higher than the one-photon absorption cross-section. This makes sense because the two-photon absorption under 785 nm fs laser excitation corresponds to one-photon absorption at 392.5 nm in the UV range, while the one-photon absorption of 785 nm is at NIR range. It is know that tissue absorption at UV range is much higher than at near IR range. Specifically previous studies have shown that the absorption coefficient of dermis at 400 nm is around ∼ 10 times higher than the absorption coefficient of dermis at 785 nm [228]. This explains why there was no tissue damage found for 785 nm CW laser irradiation. Besides, thermal diffusion was observed during the experiments, which indicates that the tissue alteration is primarily from photothermal effect. In addition, the absorption coefficient of dermis is very close to the absorption coefficient of epidermis at 400 nm [228]. Being able to damage the dermis without altering the epidermis shown in this study proved that even with similar absorption coefficient, the fs laser induced two-photon absorption based tissue alteration can be very specific. These findings indicate that the two-photon absorption based photothermolysis has great potential to be used as a novel clinical treatment tool.  7.5 Conclusion In conclusion, localized destruction of dermal fibers by two-photon absorption photothermolysis was demonstrated in ex vivo mouse skin while concurrently not damaging the overlying epidermal tissue nor the tissue below the irradiation layer. RCM and SHG+TPF images correlated well with conventional histologic examination on assessing the tissue  125  alterations. Target depths and irradiation period are important parameters that determine the amount of resulting damage. Micro-structures such as cells, hair follicles, sweat glands and sebaceous gland are very likely to have different changes when performing two-photon based photothermolysis. Therefore, future studies on optimizing laser wavelength, laser power, laser pulse width, target depth and irradiation period for different skin structures of in vivo skin should be performed. It is believed that two-photon-based light absorption could provide highly localized intradermal tissue alteration and could potentially be used for therapeutic applications in dermatology.  126  Chapter 8 Conclusion and Future Directions 8.1 Conclusion In summary, a multimodality imaging and spectroscopy system, which could noninvasively acquire co-registered reflectance confocal microscopy (RCM), two-photon fluorescence (TPF), and second harmonic generation (SHG) images simultaneously at video-rate, as well as perform image-guided region-of-interest (ROI) micro-Raman spectral measurements of human skin in vivo, has been developed. This is the first time that such an integrated system with so many different modalities has been reported. The following are the achievements of the development of the system and the novel results obtained by applying the system to skin research.  8.1.1 Confocal Raman spectroscopy The confocal Raman spectroscopy part of the system was developed and used to study both normal and neoplastic mouse skin in vivo. Different spectral patterns at different depths were found, which demonstrated the depth-resolved capability of the system. Different spectral patterns were also found in between normal and tumor-bearing skin sites. Spectral biomarkers which can be potentially used for skin cancer detection were identified and high diagnostic accuracy was achieved when classifying normal and tumor-bearing skin. 127  To improve the signal-to-noise ratio of Raman spectra, a study on fluorescence photobleaching was performed on both ex vivo mouse skin and in vivo human skin. It was found that pre-exposure to laser light significantly reduces tissue autofluorescence, but minimally affects Raman signals, allowing subsequent acquisition of high-SNR Raman spectra. It is believed that this method will benefit clinical skin Raman measurements of body sites with high autofluorescence background such as the forehead and nose, and could potentially be used for improving confocal Raman spectral quality as well.  8.1.2 Confocal Raman spectroscopy integrated with reflectance confocal microscopy imaging To achieve imaging guidance during confocal Raman spectral measurements, a reflectance confocal microscopy system was integrated with the developed confocal Raman spectroscopy system. After that, noninvasive, high-resolution, and real-time morphological and biochemical analysis can be performed on biological samples. Confocal Raman spectroscopy studies under the guidance of reflectance confocal imaging on both ex vivo and in vivo human skin were carried out. The morphological information provided by the reflectance confocal imaging enabled image-guided spectral measurements of targeted microstructures. This integrated approach is more useful for skin diagnosis and evaluation than the sole confocal Raman spectroscopy without imaging guidance.  8.1.3 Video rate multiphoton and reflectance confocal microscopy In order to obtain more information from the skin with high optical resolution, a multiphoton microscopy (MPM) instrument specially designed for in vivo dermatological use was built. The instrument is able to noninvasively acquire both TPF and SHG images of in vivo human skin at up to 27 frames per second ( f ps), which is considered as video rate, and without the use of exogenous contrast agents. Compared with other in vivo MPM instruments, which usually have an imaging rate between 0.04-2 f ps (0.5-24 second per frame) [112, 116–118, 178], the frame rate of our MPM system is more than 13 times higher. 128  Imaging at fast frame rates is critical to reduce image blurring due to patient motion and to provide practically short clinical measurement times. To facilitate more convenient in vivo optimization of the excitation wavelengths for imaging different microstructures in skin tissue, a new imaging modality was created, which integrates the SHG signal and the TPF signal into a single PMT for image acquisition (SHG+TPF imaging mode). SHG and TPF images and videos acquired at optimized wavelengths were presented showing cellular and tissue structures from the skin surface down to the reticular dermis. The capability of acquiring high-quality movies of in vivo human skin is also useful for monitoring microscale changes inside the skin. By integrating an automatic piezo scanner, the system allows us to obtain volumetric information of both ex vivo and in vivo human skin. The capability of generating 3D reconstructions of the skin images can provide more comprehensive information mimic histology and better aid skin evaluation and diagnosis. Through the studies on RCM and MPM, it was found that the two imaging modalities provide different information on in vivo skin. Therefore, with the same excitation source from the MPM part of the system, a fs reflectance confocal imaging module was built and integrated into the existing system. It has been demonstrated that, the integrated fs RCM and MPM instrument is capable of simultaneously imaging human skin in vivo at video rate with perfectly registered fs RCM, SHG, and TPF imaging channels. Complementary information of in vivo human skin were acquired from the three imaging channels. Images and videos acquired from epidermis showed that cell boundaries were clearly seen in the RCM channel, while cytoplasm and nucleus were better visualized in the TPF imaging channel, whereas in the dermis, SHG and TPF channels showed collagen bundles and elastic fibers, respectively.  8.1.4 Multimodality microscopy and spectroscopy With the four types of imaging modalities (continuous wave RCM (cwRCM), TPF, SHG and fs RCM) and the confocal Raman spectroscopy modality integrated, a study on in vivo  129  human cherry angioma lesion and surrounding normal skin to demonstrate the capability of performing accurate region-of-interest spectral measurements under the guidance of all the imaging modalities was conducted. Decent confocal Raman spectra from blood vessel of the cherry angioma, the surrounding cells and dermal fibers were obtained with the guidances of the multimodality imaging. These spectra are significantly different from each other indicating their very different biochemical compositions. The capability of realtime guidance and monitoring with the cwRCM allowed us to pre-screening the data from the confocal Raman spectral measurements and eliminate those with too much movement from the subject. This will eventually lead to better and more accurate interpretation to the spectral data. It is believed that this method will benefit for not only the confocal Raman spectral measurement, but also for all the micro level spectral measurement on in vivo subjects. Moreover, it was also able to noninvasively obtain information of the blood flow velocity and blood glucose level, which could be implemented for practical clinical uses. These results illustrated the additional applications of the developed system.  8.1.5 Two-photon absorption based photothermolysis To demonstrate more applications of the developed multimodality system, the near-infrared (NIR) fs pulses were used to perform two-photon absorption based photothermolysis on ex vivo mouse skin. RCM and SHG+TPF imaging modules were employed to monitor the high power irradiation process. Localized destruction of dermal fibers was observed while the overlying epidermal tissue and the tissue below the irradiation layer were found to be intact. RCM and SHG+TPF images correlated well with conventional histologic examination. Therefore, two-photon absorption based photothermolysis is expected to provide highly localized intradermal tissue alteration and could potentially be used for therapeutic applications in dermatology. The results also illustrated the capability of the developed system in real-time noninvasive monitoring of dynamic processes. In conclusion, the successful development of the multimodality microscopic imaging 130  and spectroscopy system as well as its applications substantiated the thesis hypothesis. The system is expected to be applied in more clinical applications after further refinements.  8.2 Future directions The capability of the developed multimodality imaging and spectroscopy system needs to be explored much more extensively. Our work to date has been proof of principle. This early work included noninvasively imaging and acquiring confocal Raman spectra of ex vivo and in vivo normal skin and very limited numbers of benign skin lesions such as nevus and angioma. In the near future, more direct comparison on different types of skin diseases with normal skin could be performed with all the imaging modalities and confocal Raman spectral measurements. A database of reflectance confocal images, multiphoton microscopic images and confocal Raman spectra of various skin diseases can be built. Analysis of these data may have significant impact on realizing noninvasive analysis and evaluation of skin and skin diseases in vivo. Moreover, studies on skin cancers with the developed system could generate spectral and imaging biomarkers for early diagnosis and significantly reduce the number of biopsies needed. Another area that should be explored is to employ this instrument for tumor margin assessment and delineation, which could potentially improve the accuracy and efficiency in Mohs micrographic skin surgery. The data that was shown in Chapter 7 on the two-photon absorption based photothermolysis in mouse skin is still very preliminary, and there is a long way to go in adapting this novel idea into clinical skin treatment. For example, different laser wavelengths, pulsewidths, irradiation durations, and irradiation power levels will very likely cause different treatment effects, therefore, studies on the relationship between these parameters and the treatment effects should be performed to find the optimized treatment protocol. Since the skin variability is another important parameter that should be considered, more research on treating different skin structures and at different treatment depths should also be con-  131  ducted. But the capability of the developed instrument allowing us to monitor the treatment in real-time can compensate for the variabilities to some extent. If the laser wavelength, pulsewidth, irradiation period and power can all be optimized for treating different skin diseases and for treating different depths, this two-photon treatment idea will move a big step towards practical clinical application. Another consideration relates to movement from the subject. So far, a magnet-based metal probe in conjunction with a double-sided tape and an arm holder to reduce the body movement has been developed, which is good enough for imaging purpose. However, for targeted two-photon based light treatment, further improvement on the design and methods for reducing involuntary body movement from the subject will be needed. One limitation of the current system is that the design of the probe permits access to a limited number of body sites, such as hands and forearm. Further modification to the probe is needed, perhaps with the development of a fiber-based probe, that will allow access to more body sites. With this enhanced functionality, one can conduct more imaging and spectral measurements on different body sites and different lesions; and compare the similarities and differences in the acquired images and spectra. Miniaturization of the developed equipment will also be beneficial for clinical use. Other applications of the developed system include applications in in vivo drug delivery, monitoring treatment process, monitoring changes in certain physiologic process, and monitoring and characterizing certain disease or tumor growth and development. The application of this system is not only limited to skin research, but it is also likely that with proper modification and improvement, the system should be able to be applied for studying other biological tissues in surface locations such as bladder, bronchus, oral cavity, and gastrointestinal tract. With the developed multimodality system, one can develop an algorithm that combines multiple signals. This may greatly enhance both the sensitivity and specificity of diagnosis, and may essentially reduce the number of biopsies and improve the diagnosic accuracy. In conclusion, the demonstrated fast imaging speed and multimodal imaging and spec132  tral measurement capabilities of this multimodality instrument are essential features for future clinical applications. The results on classifying normal and diseased skin, the novel two-photon absorption photothermolysis treatment, the capability of acquiring blood flow velocity and blood glucose level, and the unique feature of region of interest micro-Raman spectral measurement, have shown strong and versatile capabilities of this instrument in skin characterization and evaluation. With such a variety of potential applications, it is believed that this multimodality instrument could aid and have great impact in non-invasive diagnosis and in skin research, and could be used as a routine detection and evaluation device in clinic in the future.  133  Bibliography [1] W. Montagna, A.M. Kligman, and K.S. Carlisle. Atlas of normal human skin. Springer-Verlag New York, 1992. [2] G.M. James and M. Jeffery. Principles of dermatology. Elsevier Inc., 2006. [3] R. Marks, A. Plunkett, K. Merlin, and N. Jenner. Atlas of common skin diseases in Australia. Department of Dermatology, St. Vincent’s Hospital Melbourne, 1999. [4] Canadian Cancer Society’s Steering Committee on Cancer Statistics. Canadian Cancer Statistics 2012. Toronto, ON: Canadian Cancer Society, 2012. [5] American Cancer Society. Cancer Facts & Figures 2012. Atlanta: American Cancer Society, 2012. [6] J.K. Robinson. Sun exposure, sun protection, and vitamin D. JAMA: the Journal of the American Medical Association, 294(12):1541–1543, 2005. [7] R. Marks. An overview of skin cancers. Cancer, 75(S2):607–612, 1995. [8] K. Sellheyer and W.F. Bergfeld. A retrospective biopsy study of the clinical diagnostic accuracy of common skin diseases by different specialties compared with dermatology. Journal of the American Academy of Dermatology, 52(5):823–830, 2005. [9] R.W. Wood. Secret communications concerning light rays. The Journal of Physiology, 5, 1919. 134  [10] J. Margarot and P. Deveze. Aspect de quelques dermatoses en lumi`ere ultraparaviolettełnote pr´eliminaire. Bull Soc Sci Med Biol Montpellier, 6:375–378, 1925. [11] N.P. Sanchez, M.A. Pathak, S. Sato, T.B. Fitzpatrick, J.L. Sanchez, and M.C. Mihm. Melasma: a clinical, light microscopic, ultrastructural, and immunofluorescence study. Journal of the American Academy of Dermatology, 4(6):698–710, 1981. [12] OF Jilson. Wood’s light: an incredibly important diagnostic tool. Cutis; Cutaneous Medicine for the Practitioner, 28(6):620–623, 1981. [13] K.M. Halprin. Diagnosis with wood’s light. JAMA: The Journal of the American Medical Association, 199(11):841–841, 1967. [14] D.S. Rigel. Epiluminescence microscopy in clinical diagnosis of pigmented skin lesions? The Lancet, 349(9065):1566–1567, 1997. [15] R.P. Braun, H.S. Rabinovitz, M. Oliviero, A.W. Kopf, and J.H. Saurat. Dermoscopy of pigmented skin lesions. Journal of the American Academy of Dermatology, 52(1):109–121, 2005. [16] J. Mayer. Systematic review of the diagnostic accuracy of dermatoscopy in detecting malignant melanoma. Medical Journal of Australia, 167:206–210, 1997. [17] M. Binder, M. Schwarz, A. Winkler, A. Steiner, A. Kaider, K. Wolff, and H. Pehamberger. Epiluminescence microscopy: a useful tool for the diagnosis of pigmented skin lesions for formally trained dermatologists. Archives of Dermatology, 131(3):286, 1995. [18] M. Binder, M. Puespoeck-Schwarz, A. Steiner, H. Kittler, M. Muellner, K. Wolff, and H. Pehamberger. Epiluminescence microscopy of small pigmented skin lesions:  135  short-term formal training improves the diagnostic performance of dermatologists. Journal of the American Academy of Dermatology, 36(2):197–202, 1997. [19] K. Westerhoff, WH McCarthy, and SW Menzies. Increase in the sensitivity for melanoma diagnosis by primary care physicians using skin surface microscopy. British Journal of Dermatology, 143(5):1016–1020, 2000. [20] G. Argenziano, I. Zalaudek, R. Corona, F. Sera, L. Cicale, G. Petrillo, E. Ruocco, R. Hofmann-Wellenhof, and H.P. Soyer. Vascular structures in skin tumors: a dermoscopy study. Archives of Dermatology, 140(12):1485, 2004. [21] R.H. Johr. Pink lesions. Clinics in Dermatology, 20(3):289, 2002. [22] F. V´azquez-L´opez, J. Kreusch, and AA Marghoob. Dermoscopic semiology: further insights into vascular features by screening a large spectrum of nontumoral skin lesions. British Journal of Dermatology, 150(2):226–231, 2004. [23] R. Bergman, L. Sharony, D. Schapira, M.A. Nahir, and A. Balbir-Gurman. The handheld dermatoscope as a nail-fold capillaroscopic instrument. Archives of Dermatology, 139(8):1027, 2003. [24] F. V´azquez-L´opez, J.A. Manj´on-Haces, A.C. V´azquez-L´opez, N. P´erez-Oliva, et al. The hand-held dermatoscope improves the clinical evaluation of port-wine stains. Journal of the American Academy of Dermatology, 48(6):984, 2003. [25] L. Bianchi, A. Orlandi, E. Campione, C. Angeloni, A. Costanzo, LG Spagnoli, and S. Chimenti. Topical treatment of basal cell carcinoma with tazarotene: a clinicopathological study on a large series of cases. British Journal of Dermatology, 151(1):148–156, 2004. [26] F. Nachbar, W. Stolz, T. Merkle, A.B. Cognetta, T. Vogt, M. Landthaler, P. Bilek, O. Braun-Falco, and G. Plewig. The abcd rule of dermatoscopy: high prospective 136  value in the diagnosis of doubtful melanocytic skin lesions. Journal of the American Academy of Dermatology, 30(4):551–559, 1994. [27] H. Pehamberger, A. Steiner, and K. Wolff. In vivo epiluminescence microscopy of pigmented skin lesions. I. pattern analysis of pigmented skin lesions. Journal of the American Academy of Dermatology, 17(4):571–583, 1987. [28] A. Steiner, H. Pehamberger, and K. Wolff. In vivo epiluminescence microscopy of pigmented skin lesions. II. diagnosis of small pigmented skin lesions and early detection of malignant melanoma. Journal of the American Academy of Dermatology, 17(4):584–591, 1987. [29] D.A. Skoog, F.J. Holler, and S.R. Crouch. Principles of instrumental analysis. Thomson Brooks. Cole, Belmont, CA, 2007. [30] G. Zonios, J. Bykowski, and N. Kollias. Skin melanin, hemoglobin, and light scattering properties can be quantitatively assessed in vivo using diffuse reflectance spectroscopy. Journal of Investigative Dermatology, 117(6):1452–1457, 2001. [31] R. Marchesini, M. Brambilla, C. Clemente, M. Maniezzo, A.E. Sichirollo, A. Testori, D.R. Venturoli, and N. Cascinelli. In vivo spectrophotometric evaluation of neoplastic and non-neoplastic skin pigmented lesions–I. reflectance measurements. Photochemistry and Photobiology, 53(1):77–84, 1991. [32] H. Zeng and C. MacAulay. Fluorescence spectroscopy and imaging for skin cancer detection and evaluation. In Handbook of Biomedical Fluorescence, chapter 10, pages 315–360. CRC Press, New York, 2003. [33] R. Gillies, G. Zonios, R.R. Anderson, and N. Kollias. Fluorescence excitation spectroscopy provides information about human skin in vivo. Journal of Investigative Dermatology, 115(4):704–707, 2000.  137  [34] L. Brancaleon, A.J. Durkin, J.H. Tu, G. Menaker, J.D. Fallon, and N. Kollias. In vivo fluorescence spectroscopy of nonmelanoma skin cancer. Photochemistry and Photobiology, 73(2):178–183, 2001. [35] L.E. Rhodes and B.L. Diffey. Fluorescence spectroscopy: a rapid, noninvasive method for measurement of skin surface thickness of topical agents. British Journal of Dermatology, 136(1):12–17, 1997. [36] Z. Movasaghi, S. Rehman, and D.I.U. Rehman. Raman spectroscopy of biological tissues. Applied Spectroscopy Reviews, 42(5):493–541, 2007. [37] L.P. Choo-Smith, HGM Edwards, H.P. Endtz, JM Kros, F. Heule, H. Barr, JS Robinson Jr, HA Bruining, and GJ Puppels. Medical applications of Raman spectroscopy: from proof of principle to clinical implementation. Biopolymers, 67(1):1–9, 2002. [38] R. Singh and F. Riess. Sir CV Raman and the story of the Nobel Prize. Current Science, 75:965–971, 1998. [39] R. Manoharan, Y. Wang, and M.S. Feld. Histochemical analysis of biological tissues using Raman spectroscopy. Spectrochimica Acta Part A: Molecular and Biomolecular Spectroscopy, 52(2):215–249, 1996. [40] W.M. Tolles, J.W. Nibler, J.R. McDonald, and A.B. Harvey. A review of the theory and application of coherent anti-stokes Raman spectroscopy (CARS). Applied Spectroscopy, 31(4):253–271, 1977. [41] E.J. Blackie, E.C.L. Ru, and P.G. Etchegoin. Single-molecule surface-enhanced Raman spectroscopy of nonresonant molecules. Journal of the American Chemical Society, 131(40):14466–14472, 2009.  138  [42] Y. Ozaki. Medical application of Raman spectroscopy. Applied Spectroscopy Reviews, 24(3-4):259–312, 1988. [43] A. Mahadevan-Jansen and R.R. Richards-Kortum. Raman spectroscopy for the detection of cancers and precancers. Journal of Biomedical Optics, 1(01):31–70, 1996. [44] C.J. Frank, R.L. McCreery, and D.C.B. Redd. Raman spectroscopy of normal and diseased human breast tissues. Analytical Chemistry, 67(5):777–783, 1995. [45] A. Mizuno, H. Kitajima, K. Kawauchi, S. Muraishi, and Y. Ozaki. Near-infrared fourier transform Raman spectroscopic study of human brain tissues and tumours. Journal of Raman Spectroscopy, 25(1):25–29, 1994. [46] A.N.C. Anigbogu, A.C. Williams, B.W. Barry, and H.G.M. Edwards. Fourier transform Raman spectroscopy of interactions between the penetration enhancer dimethyl sulfoxide and human stratum corneum. International Journal of Pharmaceutics, 125(2):265–282, 1995. [47] A. Elhaddaoui, E. Pigorsch, A. Delacourte, and S. Turrell. Spectroscopic investigations of synthetic beta amyloid peptides of alzheimer’s disease. Journal of Molecular Structure, 347:363–369, 1995. [48] G.J. Puppels, T.C. Schut, N.M. Sijtsema, M. Grond, F. Maraboeuf, C.G. De Grauw, C.G. Figdor, and J. Greve. Development and application of Raman microspectroscopic and Raman imaging techniques for cell biological studies. Journal of Molecular Structure, 347:477–483, 1995. [49] A. Bertoluzza, C. Fagnano, A. Tinti, M.A. Morelli, M.R. Tosi, G. Maggi, and P.G. Marchetti. Raman and infrared spectroscopic study of the molecular characterization of the biocompatibility of prosthetic biomaterials. Journal of Raman Spectroscopy, 25(1):109–114, 1994. 139  [50] C.J. Frank, D.C.B. Redd, T.S. Gansler, and R.L. McCreery. Characterization of human breast biopsy specimens with near-IR Raman spectroscopy. Analytical Chemistry, 66(3):319–326, 1994. [51] A. Mahadevan-Jansen, M.F. Mitchell, N. Ramanujam, U. Utzinger, and R. Richards-Kortum. Development of a fiber optic probe to measure NIR Raman spectra of cervical tissue in vivo. Photochemistry and Photobiology, 68(3):427–431, 1998. [52] M. Gniadecka, H.C. Wulf, O.F. Nielsen, D.H. Christensen, and J. Hercogova. Distinctive molecular abnormalities in benign and malignant skin lesions: studies by Raman spectroscopy. Photochemistry and Photobiology, 66(4):418–423, 1997. [53] N. Stone, P. Stavroulaki, C. Kendall, M. Birchall, and H. Barr. Raman spectroscopy for early detection of laryngeal malignancy: preliminary results. The Laryngoscope, 110(10):1756–1763, 2000. [54] I. Siebinga, G.F.J.M. Vrensen, F.F.M. De Mul, and J. Greve. Age-related changes in local water and protein content of human eye lenses measured by Raman microspectroscopy. Experimental Eye Research, 53(2):233–239, 1991. [55] A. Molckovsky, L.M.W.K. Song, M.G. Shim, N.E. Marcon, and B.C. Wilson. Diagnostic potential of near-infrared Raman spectroscopy in the colon: Differentiating adenomatous from hyperplastic polyps. Gastrointestinal Endoscopy, 57(3):396–402, 2003. [56] T.J. R¨omer, J.F. Brennan, M. Fitzmaurice, M.L. Feldstein, G. Deinum, J.L. Myles, J.R. Kramer, R.S. Lees, and M.S. Feld. Histopathology of human coronary atherosclerosis by quantifying its chemical composition with Raman spectroscopy. Circulation, 97(9):878–885, 1998.  140  [57] J.P. Salenius, J.F. Brennan, A. Miller, Y. Wang, T. Aretz, B. Sacks, R.R. Dasari, and M.S. Feld. Biochemical composition of human peripheral arteries examined with nearinfrared Raman spectroscopy. Journal of Vascular Surgery, 27(4):710–719, 1998. [58] M.G. Shim, L.M. Wong Kee Song, N.E. Marcon, and B.C. Wilson. In vivo near-infrared Raman spectroscopy: Demonstration of feasibility during clinical gastrointestinal endoscopy. Photochemistry and Photobiology, 72(1):146–150, 2000. [59] A.S. Haka, Z. Volynskaya, J.A. Gardecki, J. Nazemi, J. Lyons, D. Hicks, M. Fitzmaurice, R.R. Dasari, J.P. Crowe, and M.S. Feld. In vivo margin assessment during partial mastectomy breast surgery using Raman spectroscopy. Cancer Research, 66(6):3317, 2006. [60] M.A. Short, S. Lam, A. McWilliams, J. Zhao, H. Lui, and H. Zeng. Development and preliminary results of an endoscopic Raman probe for potential in vivo diagnosis of lung cancers. Optics Letters, 33(7):711–713, 2008. [61] H. Lui, J. Zhao, D. McLean, and H. Zeng. Real-time Raman spectroscopy for in vivo skin cancer diagnosis. Cancer Research, 72(10):2491–2500, 2012. [62] H.P. Buschman, E.T. Marple, M.L. Wach, B. Bennett, T.C.B. Schut, H.A. Bruining, A.V. Bruschke, A. van der Laarse, and G.J. Puppels. In vivo determination of the molecular composition of artery wall by intravascular Raman spectroscopy. Analytical Chemistry, 72(16):3771–3775, 2000. [63] B.W. Barry, H.G.M. Edwards, and A.C. Williams. Fourier transform Raman and infrared vibrational study of human skin: assignment of spectral bands. Journal of Raman Spectroscopy, 23(11):641–645, 1992.  141  [64] B. Schrader. NIR FT Raman spectroscopy in medical diagnosis. Journal of Molecular Structure, 348:293–296, 1995. [65] E.E. Lawson, A.N.C. Anigbogu, A.C. Williams, B.W. Barry, and H.G.M. Edwards. Thermally induced molecular disorder in human stratum corneum lipids compared with a model phospholipid system; FT-Raman spectroscopy. Spectrochimica Acta Part A: Molecular and Biomolecular Spectroscopy, 54(3):543–558, 1998. [66] Z. Huang, H. Zeng, I. Hamzavi, D.I. McLean, and H. Lui. Rapid near-infrared Raman spectroscopy system for real-time in vivo skin measurements. Optics Letters, 26(22):1782–1784, 2001. [67] Z. Huang, H. Lui, X.K. Chen, A. Alajlan, D.I. McLean, and H. Zeng. Raman spectroscopy of in vivo cutaneous melanin. Journal of biomedical optics, 9(6):1198–1205, 2004. [68] M. Gniadecka, HC Wulf, N.N. Mortensen, O.F. Nielsen, and D.H. Christensen. Diagnosis of basal cell carcinoma by Raman spectroscopy. Journal of Raman Spectroscopy, 28(23):125–129, 1997. [69] P.J. Caspers, G.W. Lucassen, R. Wolthuis, H.A. Bruining, and G.J. Puppels. In vitro and in vivo Raman spectroscopy of human skin. Biospectroscopy, 4(S5):S31–S39, 1998. [70] M. Gniadecka, P.A. Philipsen, S. Sigurdsson, S. Wessel, O.F. Nielsen, D.H. Christensen, J. Hercogova, K. Rossen, H.K. Thomsen, R. Gniadecki, et al. Melanoma diagnosis by Raman spectroscopy and neural networks: structure alterations in proteins and lipids in intact cancer tissue. Journal of Investigative Dermatology, 122(2):443–449, 2004. [71] C.A. Lieber, S.K. Majumder, D.L. Ellis, D.D. Billheimer, and  142  A. Mahadevan-Jansen. In vivo nonmelanoma skin cancer diagnosis using Raman microspectroscopy. Lasers in Surgery and Medicine, 40(7):461–467, 2008. [72] P.J. Caspers, G.W. Lucassen, E.A. Carter, H.A. Bruining, and G.J. Puppels. In vivo confocal Raman microspectroscopy of the skin: noninvasive determination of molecular concentration profiles. Journal of Investigative Dermatology, 116(3):434–442, 2001. [73] P.D.A. Pudney, M. M´elot, P.J. Caspers, A. Van Der Pol, and G.J. Puppels. An In Vivo confocal Raman study of the delivery of trans retinol to the skin. Applied Spectroscopy, 61(8):804–811, 2007. [74] P.J. Caspers, A.C. Williams, E.A. Carter, H.G.M. Edwards, B.W. Barry, H.A. Bruining, and G.J. Puppels. Monitoring the penetration enhancer dimethyl sulfoxide in human stratum corneum in vivo by confocal Raman spectroscopy. Pharmaceutical Research, 19(10):1577–1580, 2002. [75] L. Chrit, P. Bastien, G.D. Sockalingum, D. Batisse, F. Leroy, M. Manfait, and C. Hadjur. An in vivo randomized study of human skin moisturization by a new confocal Raman fiber-optic microprobe: assessment of a glycerol-based hydration cream. Skin Pharmacology and Physiology, 19(4):207–215, 2006. [76] J.M. McCabe and K.J. Croce. Optical coherence tomography. Circulation, 126(17):2140–2143, 2012. [77] A.F. Fercher, C.K. Hitzenberger, G. Kamp, and S.Y. El-Zaiat. Measurement of intraocular distances by backscattering spectral interferometry. Optics Communications, 117(1):43–48, 1995. [78] G. Ha, M.W. Lindner, et al. Coherence radar and spectral radar − new tools for dermatological diagnosis. Journal of Biomedical Optics, 3(1):21–31, 1998.  143  [79] L.M. Smith and C.C. Dobson. Absolute displacement measurements using modulation of the spectrum of white light in a michelson interferometer. Applied Optics, 28(16):3339–3342, 1989. [80] M.A. Choma, M.V. Sarunic, C. Yang, and J.A. Izatt. Sensitivity advantage of swept source and fourier domain optical coherence tomography. Optics Express, 11(18):2183–2189, 2003. [81] S.H. Yun, G.J. Tearney, J.F. de Boer, N. Iftimia, and B.E. Bouma. High-speed optical frequency-domain imaging. Optics Express, 11(22):2953, 2003. [82] J. Welzel, E. Lankenau, R. Birngruber, and R. Engelhardt. Optical coherence tomography of the human skin. Journal of the American Academy of Dermatology, 37(6):958–963, 1997. [83] A. Pagnoni, A. Knuettel, P. Welker, M. Rist, T. Stoudemayer, L. Kolbe, I. Sadiq, and AM Kligman. Optical coherence tomography in dermatology. Skin Research and Technology, 5(2):83–87, 1999. [84] J.M. Olmedo, K.E. Warschaw, J.M. Schmitt, and D.L. Swanson. Optical coherence tomography for the characterization of basal cell carcinoma in vivo: a pilot study. Journal of the American Academy of Dermatology, 55(3):408–412, 2006. [85] J. Welzel, M. Bruhns, and H.H. Wolff. Optical coherence tomography in contact dermatitis and psoriasis. Archives of Dermatological Research, 295(2):50–55, 2003. [86] N.D. Gladkova, G.A. Petrova, N.K. Nikulin, S.G. Radenska-Lopovok, L.B. Snopova, Y.P. Chumakov, V.A. Nasonova, V.M. Gelikonov, G.V. Gelikonov, R.V. Kuranov, et al. In vivo optical coherence tomography imaging of human skin: norm and pathology. Skin Research and Technology, 6(1):6–16, 2000.  144  [87] M. Minsky. Microscopy apparatus, December 19 1957. US Patent 3,013,467. [88] M. Rajadhyaksha, M. Grossman, D. Esterowitz, R.H. Webb, and R.R. Anderson. In vivo confocal scanning laser microscopy of human skin: melanin provides strong contrast. Journal of Investigative Dermatology, 104(6):946–952, 1995. [89] M. Rajadhyaksha, R. Anderson, and R.H. Webb. Video-rate confocal scanning laser microscope for imaging human tissues in vivo. Applied Optics, 38(10):2105–2115, 1999. [90] K.B. Sung, R. Richards-Kortum, M. Follen, A. Malpica, C. Liang, and M. Descour. Fiber optic confocal reflectance microscopy: a new real-time technique to view nuclear morphology in cervical squamous epithelium in vivo. Optics Express, 11(24):3171–3181, 2003. [91] M.H. Grayson, D.D. Chaplin, I.E. Karl, and R.S. Hotchkiss. Confocal fluorescent intravital microscopy of the murine spleen. Journal of Immunological Methods, 256(1):55–63, 2001. [92] J.V. Jester, Y. Ghee Lee, J. Li, S. Chakravarti, J. Paul, W.M. Petroll, and H. Dwight Cavanagh. Measurement of corneal sublayer thickness and transparency in transgenic mice with altered corneal clarity using in vivo confocal microscopy. Vision research, 41(10):1283–1290, 2001. [93] P. Schneider, T. Foitzik, S. Kahrau, A. Podufal, and H.J. Buhr. An experimental rat model for studying pulmonary microcirculation by In Vivo videomicroscopy. Microvascular Research, 62(3):421–434, 2001. [94] A. Boyde, G. Capasso, and R.J. Unwin. Conventional and confocal epi-reflection and fluorescence microscopy of the rat kidney in vivo. Experimental Nephrology, 6(5):398–408, 1998.  145  [95] M. Rajadhyaksha, S. Gonz´alez, J.M. Zavislan, R.R. Anderson, and R.H. Webb. In vivo confocal scanning laser microscopy of human skin ii: Advances in instrumentation and comparison with histology1. Journal of Investigative Dermatology, 113(3):293–303, 1999. [96] R.G.B. Langley, M. Rajadhyaksha, P.J. Dwyer, A.J. Sober, T.J. Flotte, and R. Anderson. Confocal scanning laser microscopy of benign and malignant melanocytic skin lesions in vivo. Journal of the American Academy of Dermatology, 45(3):365–376, 2001. [97] D. Aghassi, R.R. Anderson, S. Gonz´alez, et al. Confocal laser microscopic imaging of actinic keratoses in vivo: a preliminary report. Journal of the American Academy of Dermatology, 43(1):42–48, 2000. [98] S. Astner, E. Gonzalez, A. Cheung, F. Rius-Diaz, and S. Gonz´alez. Pilot study on the sensitivity and specificity of in vivo reflectance confocal microscopy in the diagnosis of allergic contact dermatitis. Journal of the American Academy of Dermatology, 53(6):986–992, 2005. [99] S. Gonz´alez, Z. Tannous, et al. Real-time, in vivo confocal reflectance microscopy of basal cell carcinoma. Journal of the American Academy of Dermatology, 47(6):869–874, 2002. [100] M. Rajadhyaksha, G. Menaker, T. Flotte, P.J. Dwyer, and S. Gonz´alez. Confocal examination of nonmelanoma cancers in thick skin excisions to potentially guide mohs micrographic surgery without frozen histopathology. Journal of Investigative Dermatology, 117(5):1137–1143, 2001. [101] P.J. Caspers, G.W. Lucassen, and G.J. Puppels. Combined in vivo confocal Raman spectroscopy and confocal microscopy of human skin. Biophysical Journal, 85(1):572–580, 2003. 146  [102] C.L. Arrasmith, C.A. Patil, D.L. Dickensheets, and A. Mahadevan-Jansen. A MEMS based handheld confocal microscope with Raman spectroscopy for in-vivo skin cancer diagnosis. Proceedings of SPIE, 7169:71690N–1–71690N–8, 2009. ¨ [103] M. G¨oppert-Mayer. Uber elementarakte mit zwei quantenspr¨ungen. Annalen der Physik, 401(3):273–294, 1931. [104] K. K¨onig. Multiphoton microscopy in life sciences. Journal of Microscopy, 200(2):83–104, 2001. [105] K. Schenke-Layland, I. Riemann, O. Damour, U.A. Stock, and K. K¨onig. Two-photon microscopes and in vivo multiphoton tomographs powerful diagnostic tools for tissue engineering and drug delivery. Advanced Drug Delivery Reviews, 58(7):878–896, 2006. [106] S. Roth and I. Freund. Optical second-harmonic scattering in rat-tail tendon. Biopolymers, 20(6):1271–1290, 1981. [107] P.J. Campagnola, A.C. Millard, M. Terasaki, P.E. Hoppe, C.J. Malone, and W.A. Mohler. Three-dimensional high-resolution second-harmonic generation imaging of endogenous structural proteins in biological tissues. Biophysical Journal, 82:493, 2002. [108] C. Stosiek, O. Garaschuk, K. Holthoff, and A. Konnerth. In vivo two-photon calcium imaging of neuronal networks. Proceedings of the National Academy of Sciences, 100(12):7319–7324, 2003. [109] M.D. Cahalan, I. Parker, S.H. Wei, and M.J. Miller. Two-photon tissue imaging: seeing the immune system in a fresh light. Nature reviews. Immunology, 2(11):872, 2002.  147  [110] E.G. Mik, T.G. van Leeuwen, N.J. Raat, and C. Ince. Quantitative determination of localized tissue oxygen concentration in vivo by two-photon excitation phosphorescence lifetime measurements. Journal of Applied Physiology, 97(5):1962–1969, 2004. [111] D.W. Piston, B.R. Masters, and W.W. Webb. Three-dimensionally resolved NAD(P)H cellular metabolic redox imaging of the in situ cornea with two-photon excitation laser scanning microscopy. Journal of Microscopy, 178(1):20–27, 2011. [112] B.R. Masters, P.T. So, and E. Gratton. Multiphoton excitation fluorescence microscopy and spectroscopy of in vivo human skin. Biophysical Journal, 72(6):2405, 1997. [113] K. Koenig and I. Riemann. High-resolution multiphoton tomography of human skin with subcellular spatial resolution and picosecond time resolution. Journal of Biomedical Optics, 8(3):432–439, 2003. [114] K. K¨onig, A. Ehlers, F. Stracke, and I. Riemann. In vivo drug screening in human skin using femtosecond laser multiphoton tomography. Skin Pharmacology and Physiology, 19(2):78–88, 2006. [115] S.J. Lin, R. Wu, R.J., H.Y. Tan, W. Lo, W.C. Lin, T.H. Young, C.J. Hsu, J.S. Chen, S.H. Jee, and C.Y. Dong. Evaluating cutaneous photoaging by use of multiphoton fluorescence and second-harmonic generation microscopy. Optics Letters, 30(17):2275–2277, 2005. [116] S.Y. Chen, H.Y. Wu, and C.K. Sun. In vivo harmonic generation biopsy of human skin. Journal of Biomedical Optics, 14(6):060505–060505, 2009. [117] T. Yasui, Y. Takahashi, M. Ito, S. Fukushima, and T. Araki. Ex vivo and in vivo second-harmonic-generation imaging of dermal collagen fiber in skin: comparison  148  of imaging characteristics between mode-locked Cr: forsterite and Ti: sapphire lasers. Applied Optics, 48(10):D88–D95, 2009. [118] A.N. Bader, A.M. Pena, C. Johan van Voskuilen, J.A. Palero, F. Leroy, A. Colonna, and H.C. Gerritsen. Fast nonlinear spectral microscopy of in vivo human skin. Biomedical Optics Express, 2(2):365–373, 2011. [119] T.H. Tsai, S.H. Jee, C.Y. Dong, and S.J. Lin. Multiphoton microscopy in dermatological imaging. Journal of Dermatological Science, 56(1):1, 2009. [120] J. Paoli, M. Smedh, A.M. Wennberg, and M.B. Ericson. Multiphoton laser scanning microscopy on non-melanoma skin cancer: morphologic features for future non-invasive diagnostics. Journal of Investigative Dermatology, 128(5):1248–1255, 2007. [121] S.J. Lin, S.H. Jee, C.J. Kuo, R. Wu Jr, W.C. Lin, J.S. Chen, Y.H. Liao, C.J. Hsu, T.F. Tsai, Y.F. Chen, et al. Discrimination of basal cell carcinoma from normal dermal stroma by quantitative multiphoton imaging. Optics Letters, 31(18):2756–2758, 2006. [122] E. Dimitrow, M. Ziemer, M.J. Koehler, J. Norgauer, K. K¨onig, P. Elsner, and M. Kaatz. Sensitivity and specificity of multiphoton laser tomography for in vivo and ex vivo diagnosis of malignant melanoma. Journal of Investigative Dermatology, 129(7):1752–1758, 2009. [123] F. Stracke, B. Weiss, C.M. Lehr, K. K¨onig, U.F. Schaefer, and M. Schneider. Multiphoton microscopy for the investigation of dermal penetration of nanoparticle-borne drugs. Journal of Investigative Dermatology, 126(10):2224–2233, 2006. [124] R.S. Becker. Theory and interpretation of fluorescence and phosphorescence. Wiley Interscience New York, NY, 1969. 149  [125] J.D. Winefordner. Practical fluorescence: theory, methods, and techniques. Journal of Chemical Education, 51(1):48, 1974. [126] J.R. Lakowicz. Principles of fluorescence spectroscopy, volume 1. Springer, 2006. [127] J. White, I. Horowitz, and D. Dederich. American National Standard for Safe Use of Lasers ANSI Z 136.12007. Laser Institute of America: Orlando, 2007. [128] S.W. Smith et al. The scientist and engineer’s guide to digital signal processing. California Technical Pub. San Diego, 1997. [129] N.J. Everall. Modeling and measuring the effect of refraction on the depth resolution of confocal raman microscopy. Applied Spectroscopy, 54(6):773–782, 2000. [130] J. Zhao, H. Lui, D.I. McLean, and H. Zeng. Automated autofluorescence background subtraction algorithm for biomedical Raman spectroscopy. Applied Spectroscopy, 61(11):1225–1232, 2007. [131] M.J. Goetz Jr, GL Cot´e, R. Erckens, W. March, and M. Motamedi. Application of a multivariate technique to Raman spectra for quantification of body chemicals. Biomedical Engineering, IEEE Transactions on, 42(7):728–731, 1995. [132] P.A. Lachenbruch and M.R. Mickey. Estimation of error rates in discriminant analysis. Technometrics, pages 1–11, 1968. [133] J.W. Chan, D.S. Taylor, T. Zwerdling, S.M. Lane, K. Ihara, and T. Huser. Micro-Raman spectroscopy detects individual neoplastic and normal hematopoietic cells. Biophysical Journal, 90(2):648–656, 2006. [134] R. Malini, K. Venkatakrishna, J. Kurien, K. M Pai, L. Rao, VB Kartha, and C.M. Krishna. Discrimination of normal, inflammatory, premalignant, and malignant oral tissue: a Raman spectroscopy study. Biopolymers, 81(3):179–193, 2006. 150  [135] A. Robichaux Viehoever, D. Anderson, D. Jansen, and A. Mahadevan-Jansen. Organotypic raft cultures as an effective In Vitro tool for understanding Raman spectral analysis of tissue. Photochemistry and Photobiology, 78(5):517–524, 2003. [136] C. Krafft, L. Neudert, T. Simat, R. Salzer, et al. Near infrared Raman spectra of human brain lipids. Spectrochimica acta. Part A, Molecular and biomolecular spectroscopy, 61(7):1529, 2005. [137] N. Stone, C. Kendall, J. Smith, P. Crow, and H. Barr. Raman spectroscopy for identification of epithelial cancers. Faraday Discuss., 126:141–157, 2003. [138] W.T. Cheng, M.T. Liu, H.N. Liu, and S.Y. Lin. Micro-Raman spectroscopy used to identify and grade human skin pilomatrixoma. Microscopy Research and Technique, 68(2):75–79, 2005. [139] Z. Movasaghi, S. Rehman, and D.I. ur Rehman. Fourier transform infrared (FTIR) spectroscopy of biological tissues. Applied Spectroscopy Reviews, 43(2):134–179, 2008. [140] I. Notingher, C. Green, C. Dyer, E. Perkins, N. Hopkins, C. Lindsay, and L.L. Hench. Discrimination between ricin and sulphur mustard toxicity in vitro using Raman spectroscopy. Journal of The Royal Society Interface, 1(1):79–90, 2004. [141] Z. Huang, A. McWilliams, H. Lui, D.I. McLean, S. Lam, and H. Zeng. Near-infrared Raman spectroscopy for optical diagnosis of lung cancer. International Journal of Cancer, 107(6):1047–1052, 2003. [142] E. Katainen, M. Elomaa, U.M. Laakkonen, E. Sippola, P. Niemel¨a, J. Suhonen, and K. J¨arvinen. Quantification of the amphetamine content in seized street samples by Raman spectroscopy. Journal of Forensic Sciences, 52(1):88–92, 2007.  151  [143] M. De Heller-Milev, M. Huber, R. Panizzon, and D. Hohl. Expression of small proline rich proteins in neoplastic and inflammatory skin diseases. British Journal of Dermatology, 143(4):733–740, 2000. [144] T. Kartasova, GN Van Muijen, H. van Pelt-Heerschap, and P. Van de Putte. Novel protein in human epidermal keratinocytes: regulation of expression during differentiation. Molecular and Cellular Biology, 8(5):2204–2210, 1988. [145] T. Kartasova and P. van de Putte. Isolation, characterization, and UV-stimulated expression of two families of genes encoding polypeptides of related structure in human epidermal keratinocytes. Molecular and Cellular Biology, 8(5):2195–2203, 1988. [146] S. Gibbs, R. Fijneman, J. Wiegant, A.G. van Kessel, P. van de Putte, and C. Backendorf. Molecular characterization and evolution of the SPRR family of keratinocyte differentiation markers encoding small proline-rich proteins. Genomics, 16(3):630–637, 1993. [147] H. Koizumi, T. Kartasova, H. Tanaka, A. Ohkawara, and T. Kuroki. Differentiation-associated localization of small prolne-rich rotein in normal and diseased human skin. British Journal of Dermatology, 134(4):686–692, 1996. [148] J. Brod. Characterization and physiological role of epidermal lipids. International Journal of Dermatology, 30(2):84–90, 1991. [149] AM Engelbrecht, L. Louw, and F. Cloete. Comparison of the fatty acid compositions in intraepithelial and infiltrating lesions of the cervix: part II, free fatty acid profiles. Prostaglandins, Leukotrienes and Essential Fatty Acids, 59(4):253–257, 1998. [150] G. Mamalakis, A. Kafatos, N. Kalogeropoulos, N. Andrikopoulos, G. Daskalopulos, and A. Kranidis. Prostate cancer vs hyperplasia: relationships 152  with prostatic and adipose tissue fatty acid composition. Prostaglandins, Leukotrienes and Essential Fatty Acids, 66(5-6):467–477, 2002. [151] M. Pandey, A.K. Khatri, SS Dubey, A. Gautam, and V.K. Shukla. Erythrocyte membrane stearic to oleic acid ratio in carcinoma of the gallbladder: a preliminary study. European Journal of Surgical Oncology, 24(1):43–46, 1998. [152] S.B. Kelly, J. Miller, C.B. Wood, R.C.N. Williamson, and N.A. Habib. Erythrocyte stearic acid desaturation in patients with colorectal carcinoma. Diseases of the Colon & Rectum, 33(12):1026–1030, 1990. [153] N. Lawson, AJ Taylor, A. Manche, D. Watson, and HI Pandov. Inadequacy of oleic acid in erythrocytes as a marker of malignancies. British Medical Journal, 294(6574):769–769, 1987. [154] J. Guo, W. Cai, B. Du, M. Qian, and Z. Sun. Raman spectroscopic investigation on the interaction of malignanthepatocytes with doxorubicin. Biophysical Chemistry, 140(1-3):57–61, 2009. [155] C.A. Elstad, G.G. Meadows, and R.M. Abdallah. Specificity of the suppression of metastatic phenotype by tyrosine and phenylalanine restriction. Clinical and Experimental Metastasis, 8(5):393–416, 1990. [156] R. Peters, J. Peters, K.H. Tews, and W. B¨ahr. A microfluorimetric study of translational diffusion in erythrocyte membranes. Biochimica et Biophysica Acta (BBA)-Biomembranes, 367(3):282–294, 1974. [157] D. Axelrod, D.E. Koppel, J. Schlessinger, E. Elson, and W.W. Webb. Mobility measurement by analysis of fluorescence photobleaching recovery kinetics. Biophysical Journal, 16(9):1055, 1976. [158] M.A. Mycek and B.W. Pogue. Handbook of Biomedical Fluorescence. CRC, 2003. 153  [159] H. Zeng, C. MacAulay, D.I. McLean, B. Palcic, and H. Lui. The dynamics of laser-induced changes in human skin autofluorescence – experimental measurements and theoretical modeling. Photochemistry and Photobiology, 68(2):227–236, 1998. [160] J. Pawley. Handbook of Biological Confocal Microscopy. Springer, 2006. [161] T. Bernas, M. Zarebski, R.R. Cook, and J.W. Dobrucki. Minimizing photobleaching during confocal microscopy of fluorescent probes bound to chromatin: role of anoxia and photon flux. Journal of Microscopy, 215(3):281–296, 2004. [162] J. Zhao, H. Lui, D.I. McLean, and H. Zeng. Integrated real-time Raman system for clinical in vivo skin analysis. Skin Research and Technology, 14(4):484–492, 2008. [163] H. Wang, N. Huang, J. Zhao, H. Lui, M. Korbelik, and H. Zeng. Depth-resolved in vivo micro-Raman spectroscopy of a murine skin tumor model reveals cancer-specific spectral biomarkers. Journal of Raman Spectroscopy, 42(2):160–166, 2011. [164] C. Lieber and A. Mahadevan-Jansen. Development of a handheld raman microspectrometer for clinical dermatologic applications. Optics Express, 15(19):11874–11882, 2007. [165] G.J. Puppels, F.F.M. De Mul, C. Otto, J. Greve, M. Robert-Nicoud, D.J. Arndt-Jovin, and T.M. Jovin. Studying single living cells and chromosomes by confocal Raman microspectroscopy. Nature, 347:301–303, 1990. [166] J. Choi, J. Choo, H. Chung, D.G. Gweon, J. Park, H.J. Kim, S. Park, and C.H. Oh. Direct observation of spectral differences between normal and basal cell carcinoma (BCC) tissues using confocal Raman microscopy. Biopolymers, 77(5):264–272, 2005.  154  [167] T. Wilson. Confocal microscopy. Academic Press: London, etc, 426:1–64, 1990. [168] B.R. Masters and P.T.C. So. Confocal microscopy and multi-photon excitation microscopy of human skin in vivo. Optics Express, 8(1):2–10, 2001. [169] S. Seidenari, F. Arginelli, S. Bassoli, J. Cautela, P.M.W. French, M. Guanti, D. Guardoli, K. K¨onig, C. Talbot, and C. Dunsby. Multiphoton laser microscopy and fluorescence lifetime imaging for the evaluation of the skin. Dermatology Research and Practice, 2012:1–8, 2011. [170] P.J. Campagnola and L.M. Loew. Second-harmonic imaging microscopy for visualizing biomolecular arrays in cells, tissues and organisms. Nature Biotechnology, 21(11):1356–1360, 2003. [171] W.R. Zipfel, R.M. Williams, R. Christie, A.Y. Nikitin, B.T. Hyman, and W.W. Webb. Live tissue intrinsic emission microscopy using multiphoton-excited native fluorescence and second harmonic generation. Proceedings of the National Academy of Sciences, 100(12):7075–7080, 2003. [172] K. K¨onig. Clinical multiphoton tomography. Journal of Biophotonics, 1(1):13–23, 2008. [173] D.A. Dombeck, K.A. Kasischke, H.D. Vishwasrao, M. Ingelsson, B.T. Hyman, and W.W. Webb. Uniform polarity microtubule assemblies imaged in native brain tissue by second-harmonic generation microscopy. Proceedings of the National Academy of Sciences, 100(12):7081–7086, 2003. [174] W. Mohler, A.C. Millard, and P.J. Campagnola. Second harmonic generation imaging of endogenous structural proteins. Methods, 29(1):97–109, 2003. [175] E. Brown, T. McKee, et al. Dynamic imaging of collagen and its modulation in  155  tumors in vivo using second-harmonic generation. Nature Medicine, 9(6):796–800, 2003. [176] G. Cox, E. Kable, A. Jones, I. Fraser, F. Manconi, and M.D. Gorrell. 3-dimensional imaging of collagen using second harmonic generation. Journal of Structural Biology, 141(1):53–62, 2003. [177] W. Denk, J.H. Strickler, W.W. Webb, et al. Two-photon laser scanning fluorescence microscopy. Science (New York, NY), 248(4951):73, 1990. [178] K. K¨onig, M. Speicher, M.J. K¨ohler, R. Scharenberg, and M. Kaatz. Clinical application of multiphoton tomography in combination with high-frequency ultrasound for evaluation of skin diseases. Journal of Biophotonics, 3(12):759–773, 2010. [179] G.Y. Fan, H. Fujisaki, A. Miyawaki, RK Tsay, R.Y. Tsien, and M.H. Ellisman. Video-rate scanning two-photon excitation fluorescence microscopy and ratio imaging with cameleons. Biophysical Journal, 76(5):2412, 1999. [180] B. Yu, K.H. Kim, P.T.C. So, D. Blankschtein, and R. Langer. Visualization of oleic acid-induced transdermal diffusion pathways using two-photon fluorescence microscopy. Journal of Investigative Dermatology, 120(3):448–455, 2003. [181] S.E.G. Fligiel, J. Varani, S.C. Datta, S. Kang, G.J. Fisher, and J.J. Voorhees. Collagen degradation in aged/photodamaged skin in vivo and after exposure to matrix metalloproteinase-1 in vitro. Journal of Investigative Dermatology, 120(5):842–848, 2003. [182] A. Karotki, M. Khurana, J.R. Lepock, and B.C. Wilson. Simultaneous two-photon excitation of photofrin in relation to photodynamic therapy. Photochemistry and photobiology, 82(2):443–452, 2006.  156  [183] V.V. Ghukasyan and F.J. Kao. Monitoring cellular metabolism with fluorescence lifetime of reduced nicotinamide adenine dinucleotide? The Journal of Physical Chemistry C, 113(27):11532–11540, 2009. [184] M.C. Skala, K.M. Riching, D.K. Bird, A. Gendron-Fitzpatrick, J. Eickhoff, K.W. Eliceiri, P.J. Keely, and N. Ramanujam. In vivo multiphoton fluorescence lifetime imaging of protein-bound and free nicotinamide adenine dinucleotide in normal and precancerous epithelia. Journal of Biomedical Optics, 12(2):024014–024014, 2007. [185] E. Dimitrow, I. Riemann, A. Ehlers, M.J. Koehler, J. Norgauer, P. Elsner, K. K¨onig, and M. Kaatz. Spectral fluorescence lifetime detection and selective melanin imaging by multiphoton laser tomography for melanoma diagnosis. Experimental Dermatology, 18(6):509–515, 2009. [186] L.H. Laiho, S. Pelet, T.M. Hancewicz, P.D. Kaplan, and P.T.C. So. Two-photon 3-D mapping of ex vivo human skin endogenous fluorescence species based on fluorescence emission spectra. Journal of Biomedical Optics, 10(2):024016–024016, 2005. [187] M.J. Koehler, K. K¨onig, P. Elsner, R. B¨uckle, and M. Kaatz. In vivo assessment of human skin aging by multiphoton laser scanning tomography. Optics Letters, 31(19):2879–2881, 2006. [188] K.E. Kadler, D.F. Holmes, J.A. Trotter, and J.A. Chapman. Collagen fibril formation. Biochemical Journal, 316(Pt 1):1, 1996. [189] I. Freund, M. Deutsch, and A. Sprecher. Connective tissue polarity. Optical second-harmonic microscopy, crossed-beam summation, and small-angle scattering in rat-tail tendon. Biophysical Journal, 50(4):693–712, 1986. [190] P. Calzavara-Pinton, C. Longo, M. Venturini, R. Sala, and G. Pellacani. Reflectance  157  confocal microscopy for In Vivo skin imaging. Photochemistry and Photobiology, 84(6):1421–1430, 2008. [191] R.J.D. Tilley. Crystals and crystal structures. Wiley, 2006. [192] W.L. Chen, C.K. Chou, M.G. Lin, Y.F. Chen, S.H. Jee, H.Y. Tan, T.H. Tsai, K.H. Kim, D. Kim, P.T.C. So, et al. Single-wavelength reflected confocal and multiphoton microscopy for tissue imaging. Journal of Biomedical Optics, 14:054026, 2009. [193] M.G. Lin, W.L. Chen, W. Lo, H.Y. Tan, T.H. Tsai, S.H. Jee, S.J. Lin, and C.Y. Dong. Three-dimensional skin imaging using the combination of reflected confocal and multiphoton microscopy. Proceedings of SPIE, 6424:64240E, 2007. [194] S.H. Choi, W.H. Kim, Y.J. Lee, H. Lee, W.J. Lee, J.D. Yang, J.W. Shim, and J.W. Kim. Visualization of epidermis and dermal cells in ex vivo human skin using the confocal and two-photon microscopy. Journal of the Optical Society of Korea, 15(1):61–67, 2011. [195] M.J. Koehler, M. Speicher, S. Lange-Asschenfeldt, E. Stockfleth, S. Metz, P. Elsner, M. Kaatz, and K. K¨onig. Clinical application of multiphoton tomography in combination with confocal laser scanning microscopy for in vivo evaluation of skin diseases. Experimental Dermatology, 20:589–594, 2011. [196] S. Laffray, S. Pag`es, H. Dufour, P. De Koninck, Y. De Koninck, and D. Cˆot´e. Adaptive movement compensation for In Vivo imaging of fast cellular dynamics within a moving tissue. PloS One, 6(5):e19928, 2011. [197] A. Lee, H. Wang, Y. Yu, S. Tang, J. Zhao, H. Lui, D.I. McLean, and H. Zeng. In vivo video rate multiphoton microscopy imaging of human skin. Optics Letters, 36(15):2865–2867, 2011.  158  [198] W.L. Chen, Y. Sun, W. Lo, H.Y. Tan, and C.Y. Dong. Combination of multiphoton and reflective confocal imaging of cornea. Microscopy Research and Technique, 71(2):83–85, 2008. [199] S.M. Zhuo, J.X. Chen, X.S. Jiang, K.C. Lu, and S.S. Xie. Imaging rat esophagus using combination of reflectance confocal and multiphoton microscopy. Laser Physics Letters, 5(8):614–618, 2008. [200] W.L. Chen, W. Lo, Y. Sun, S.J. Lin, H.Y. Tan, and C.Y. Dong. The combination of multiphoton and reflected confocal microscopy for cornea imaging. Proceedings of SPIE, 6138:61380M, 2006. [201] R.F.M. Hendriks and G.W. Lucassen. Two-photon fluorescence and confocal video microscopy of in-vivo human skin. Proceedings of SPIE, 4262:287, 2001. [202] M.A. Luedtke, E. Papazoglou, M. Neidrauer, and N. Kollias. Wavelength effects on contrast observed with reflectance in vivo confocal laser scanning microscopy. Skin Research and Technology, 15(4):482–488, 2009. [203] S.L. Mera, C.R. Lovell, R.R. Jones, and J.D. Davies. Elastic fibres in normal and sun-damaged skin: an immunohistochemical study. British Journal of Dermatology, 117(1):21–27, 2006. [204] H. Bouissou, M.T. Pieraggi, M. Julian, and T. Savait. The elastic tissue of the skin. International Journal of Dermatology, 27(5):327–335, 2007. [205] W. Montagna, S. Kirchner, and K. Carlisle. Histology of sun-damaged human skin. Journal of the American Academy of Dermatology, 21(5):907–918, 1989. [206] N.J.C. Bauer, F. Hendrikse, and W.F. March. In vivo confocal Raman spectroscopy of the human cornea. Cornea, 18(4):483–488, 1999.  159  [207] M. Egawa, T. Hirao, and M. Takahashi. In vivo estimation of stratum corneum thickness from water concentration profiles obtained with Raman spectroscopy. Acta Dermato-Venereologica, 87(1):4–8, 2007. [208] K.E. Shafer-Peltier, A.S. Haka, M. Fitzmaurice, J. Crowe, J. Myles, R.R. Dasari, and M.S. Feld. Raman microspectroscopic model of human breast tissue: implications for breast cancer diagnosis in vivo. Journal of Raman Spectroscopy, 33(7):552–563, 2002. [209] S. Mordechai, R.K. Sahu, Z. Hammody, S. Mark, K. Kantarovich, H. Guterman, A. Podshyvalov, J. Goldstein, and S. Argov. Possible common biomarkers from ftir microspectroscopy of cervical cancer and melanoma. Journal of Microscopy, 215(1):86–91, 2004. [210] G. Zhang, D.J. Moore, C.R. Flach, and R. Mendelsohn. Vibrational microscopy and imaging of skin: from single cells to intact tissue. Analytical and Bioanalytical Chemistry, 387(5):1591–1599, 2007. [211] H. Wang, A. Lee, Z. Frehlick, H. Lui, D.I. McLean, S. Tang, and H. Zeng. Perfectly registered multiphoton and reflectance confocal video rate imaging of in vivo human skin. Journal of Biophotonics, 2012. [212] J. Binoy, J.P. Abraham, I.H. Joe, VS Jayakumar, GR Pettit, and O.F. Nielsen. NIR-FT Raman and FT-IR spectral studies and ab initio calculations of the anti-cancer drug combretastatin-A4. Journal of Raman Spectroscopy, 35(11):939–946, 2004. [213] J.L. Deng, Q. Wei, M.H. Zhang, Y.Z. Wang, and Y.Q. Li. Study of the effect of alcohol on single human red blood cells using near-infrared laser tweezers Raman spectroscopy. Journal of Raman Spectroscopy, 36(3):257–261, 2005.  160  [214] D. Naumaim. Infrared and NIR Raman spectroscopy in medical microbiology. Proceedings of SPIE, 3257:245–257, 1998. [215] D.P. Lau, Z. Huang, H. Lui, C.S. Man, K. Berean, M.D. Morrison, and H. Zeng. Raman spectroscopy for optical diagnosis in normal and cancerous tissue of the nasopharynxpreliminary findings. Lasers in Surgery and Medicine, 32(3):210–214, 2003. [216] A.J. Berger, I. Itzkan, and M.S. Feld. Feasibility of measuring blood glucose concentration by near-infrared Raman spectroscopy. Spectrochimica Acta Part A: Molecular and Biomolecular Spectroscopy, 53(2):287–292, 1997. [217] A.M.K. Enejder, T.G. Scecina, J. Oh, M. Hunter, W.C. Shih, S. Sasic, G.L. Horowitz, and M.S. Feld. Raman spectroscopy for noninvasive glucose measurements. Journal of Biomedical Optics, 10(3):031114–031114, 2005. [218] R.R. Anderson, J.A. Parrish, et al. Selective photothermolysis: precise microsurgery by selective absorption of pulsed radiation. Science, 220(4596):524–527, 1983. [219] D.H. Sliney et al. Laser-tissue interactions. Clinics in Chest Medicine, 6(2):203, 1985. [220] E.L. Tanzi, J.R. Lupton, T.S. Alster, et al. Lasers in dermatology: four decades of progress. Journal of the American Academy of Dermatology, 49(1):1, 2003. [221] K. Jang, E.C. Chung, J.H. Choi, K.J. Sung, K.C. Moon, and J.K. Koh. Successful removal of freckles in asian skin with a Q-switched alexandrite laser. Dermatologic Surgery, 26(3):231–234, 2001. [222] B.D. Zelickson, S.L. Kilmer, E. Bernstein, V.A. Chotzen, J. Dock, D. Mehregan,  161  and C. Coles. Pulsed dye laser therapy for sun damaged skin. Lasers in Surgery and Medicine, 25(3):229–236, 2000. [223] J.D. Bhawalkar, N.D. Kumar, C.F. Zhao, and P.N. Prasad. Two-photon photodynamic therapy. Journal of Clinical Laser Medicine & Surgery, 15(5):201–204, 1997. [224] H. Sun, M. Han, M.H. Niemz, and J.F. Bille. Femtosecond laser corneal ablation threshold: dependence on tissue depth and laser pulse width. Lasers in Surgery and Medicine, 39(8):654–658, 2007. [225] W. Kautek, S. Mitterer, J. Kr¨uger, W. Husinsky, and G. Grabner. Femtosecond-pulse laser ablation of human corneas. Applied Physics A: Materials Science & Processing, 58(5):513–518, 1994. [226] T.A. Theodossiou, C. Thrasivoulou, C. Ekwobi, and D.L. Becker. Second harmonic generation confocal microscopy of collagen type I from rat tendon cryosections. Biophysical Journal, 91(12):4665–4677, 2006. [227] A.J. Walsh, D.B. Masters, E.D. Jansen, A.J. Welch, and A. Mahadevan-Jansen. The effect of temperature on the autofluorescence of scattering and non-scattering tissue. Lasers in Surgery and Medicine, 44(9):712–718, 2012. [228] E. Salomatina, B. Jiang, J. Novak, and A.N. Yaroslavsky. Optical properties of normal and cancerous human skin in the visible and near-infrared spectral range. Journal of Biomedical Optics, 11(6):064026–064026, 2006.  162  Appendix Captions for Supplementary Media Supplementary Media 1: In vivo integrated SHG+TPF video from epidermis (oblique orientation) of the dorsal forearm of a 24-year-old Asian female. Excitation wavelength = 730 nm. FOV = 200μ m×200μ m. Resolution = 256pixels×256pixels. Frame scan rate = 27 f ps. Frame capture/display rate = 24 f ps. At this wavelength, the cells of the epidermis are well visualized from the stratum corneum (SC) down to the stratum basale (SB). Supplementary Media 2: In vivo TPF video from epidermis (oblique orientation) of the dorsal forearm of a 25-year-old Asian female increasing in depth from near the surface to near the dermal/epidermal boundary. Excitation wavelength = 730 nm. FOV = 200μ m×200μ m. Resolution = 256pixels×256pixels. Frame scan rate = 15 f ps. Frame capture/display rate = 12 f ps. At this wavelength, the cells of the epidermis are well visualized from the SC down to the SB. Supplementary Media 3: In vivo integrated SHG+TPF video from the epidermal ridges and the papillary dermis of the dorsal forearm of a 28-year-old Asian male. The imaging depth increases from the dermal/epidermal boundary. Excitation wavelength = 900 nm. FOV = 200μ m×200μ m. Resolution = 256pixels×256pixels. Frame scan rate = 15 f ps. Frame capture/display rate = 12 f ps. At this wavelength, the cells of the SB and collagen and elastin fibers are observed. Supplementary Media 4: In vivo false color overlay of SHG (green) and TPF (red) video from the reticular dermis of the dorsal forearm of a 63-year-old Caucasian male. Excitation wavelength = 880 nm. FOV = 200μ m×200μ m. Resolution = 256pixels×256pixels. Frame scan rate = 15 f ps. Frame capture/display rate = 12 f ps. Collagen (SHG) and elastin (TPF) fibers are clearly seen. Supplementary Media 5: Video of 3D reconstruction of ex vivo human cheek skin from XY-Z scan. Acquisition time: 2 min; depth: 50 μ m; stepsize: 0.5 μ m; FOV: 150μ m×150μ m; 163  laser wavelength: 730 nm; and laser power: 30 mW . Supplementary Media 6: Video of 3D reconstruction of ex vivo human eyelid skin from XY-Z scan. Acquisition time: 2 min; depth: 130 μ m; stepsize: 1 μ m; FOV: 150μ m×150μ m; laser wavelength: 730 nm; and laser power: 30 mW . Supplementary Media 7: Video of 3D reconstruction of in vivo human forearm skin from XY-Z scan. Acquisition time: 2.5 min; depth: 130 μ m; stepsize: 1 μ m; FOV: 150μ m×150μ m; laser wavelength: 730 nm; and laser power: 30 mW . Supplementary Media 8: Video of 3D reconstruction of in vivo human sweat duct from XY-Z scan. Acquisition time: 2.5 min; depth: 130 μ m; stepsize: 1 μ m; FOV: 150μ m×150μ m; laser wavelength: 730 nm; and laser power: 30 mW . Supplementary Media 9: In vivo false color overlay of SHG+TPF (green) and RCM (red) video from the stratum basale (SB) of the dorsal forearm of a Caucasian male in his early 50’s. Excitation wavelength = 730 nm. FOV = 170μ m × 170μ m. Resolution = 256pixels×256pixels. Frame rate = 15 f ps. Supplementary Media 10: In vivo false color overlay of SHG+TPF (green) and RCM (red) video from the stratum granulosum (SG) of the dorsal forearm of a 41-year-old Asian male. Excitation wavelength = 730 nm. FOV = 150μ m×150μ m. Resolution = 256pixels×256pixels. Frame rate = 27 f ps. Supplementary Media 11: In vivo false color overlay of RCM (red), TPF (green), and SHG (blue) video from the reticular dermis (RD) of the ventral forearm of a 64-year-old Caucasian male. Excitation wavelength = 880 nm. FOV = 150μ m × 150μ m. Resolution = 256pixels×256pixels. Frame rate = 15 f ps. Supplementary Media 12: In vivo integrated SHG+TPF video from the stratum granulosum to reticular dermis of the dorsal forearm of a 23-year-old Asian male (optic compressor introduced into the system). Excitation wavelength =750 nm. FOV = 150μ m×150μ m. Resolution = 256pixels×256pixels. Frame rate = 15 f ps. Supplementary Media 13: In vivo cwRCM video demonstrating the selection of an ROI  164  for spectral measurement of a blood vessel inside a cherry angioma lesion on the dorsal upper arm of a 48-year-old volunteer. Laser wavelength: 785 nm. Irradiation power: 27 mw. The FOV is changed from 300μ m×300μ m down to 80μ m×80μ m. Supplementary Media 14: In vivo cwRCM video recorded during a confocal Raman spectral measurement of a blood vessel in a cherry angioma lesion on the dorsal upper arm of a 48-year-old volunteer. Laser wavelength: 785 nm. Irradiation power: 27 mw. FOV = 100μ m×100μ m. Supplementary Media 15: RCM video showing the two photon treatment process of ex vivo mouse skin. Laser wavelength: 785 nm, treatment power: 75 mw, treatment depth: 20  μ m , treatment period: 0.5 min, treatment area: 70μ m × 70μ m. Supplementary Media 16: SHG+TPF video showing the two photon treatment process of ex vivo mouse skin. Laser wavelength: 785 nm, treatment power: 75 mw, treatment depth: 20 μ m , treatment period: 0.5 min, treatment area: 70μ m × 70μ m.  165  

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
http://iiif.library.ubc.ca/presentation/dsp.24.1-0073699/manifest

Comment

Related Items