Optical coherence tomography for optical biopsy of axillary lymph nodes involved in breast cancer metastasis Loretta Scolaro BE(Hons), BSc This thesis is presented for the degree of Doctor of Philosophy of The University of Western Australia School of Electrical, Electronic & Computer Engineering. January 2014
54
Embed
Optical coherence tomography for optical biopsy of ...
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Optical coherence tomography for �“optical biopsy�” of axillary lymph nodes
involved in breast cancer metastasis
Loretta Scolaro BE(Hons), BSc
This thesis is presented for the degree of
Doctor of Philosophy
of The University of Western Australia
School of Electrical, Electronic & Computer Engineering.
January 2014
ii
iii
Abstract One of the pathways for the spread of metastatic breast cancer throughout the body is the
lymphatic system. Accurate determination of the presence and extent of metastatic deposits in
lymph nodes of the axilla (staging) is critical to the management of the disease. Currently,
lymph nodes are assessed by microscopic examination after surgical excision. However, this
can mean removal of healthy lymph nodes without metastatic disease, and may result in
unnecessary lymphoedema; that is, swelling due to accumulation of tissue fluid as a result of
disruption to normal lymphatic pathway function. This thesis presents an investigation into a
possible optical method of lymph node assessment based on optical coherence tomography
(OCT) that has the potential to provide in situ “optical biopsy” of lymph node involvement.
In situ biopsy could reduce the rate of unnecessary lymphoedema and improve efficiency of
node assessment.
In this thesis, an investigation of the baseline appearance of the morphology of human
axillary lymph nodes imaged using OCT is first presented. We assess OCT images of excised
human axillary lymph nodes; both healthy nodes and nodes involved in metastatic spread of
breast cancer are imaged using a benchtop OCT system. We compare OCT images with the
structural and cellular composition identified using the gold standard technique of
histopathology. The results identify a higher OCT signal from metastatic deposits and show
that OCT can successfully image the micro-architecture in lymph nodes.
It was evident from this study, however, that the diagnostic capability of OCT requires
improvement. The native contrast of intensity-based OCT is not, in itself, sufficient to
differentiate metastatic deposits in involved lymph nodes, primarily because interpretation of
such contrast is ambiguous and subject to artefacts. One of the most problematic artefacts
occurs as a result of the variable attenuation of light in tissue. In this thesis, we therefore
present a method for processing the OCT data to overcome these ambiguities. The method
exploits the attenuation of light in tissue to provide a quantitative measure of contrast. It
utilises a single-scattering model to describe the OCT reflectance from tissue and extract
optical attenuation coefficients. The attenuation coefficient is an optical property of tissue
that is determined by its unique cellular and structural composition and is, therefore, expected
to vary between healthy and malignant tissue. We demonstrate improved differentiation of
healthy and abnormal lymph node tissue types based on measurement of this coefficient and
iv
use a visual representation to interpret the measured values by mapping them into en face
parametric images.
A second improvement we propose for enabling OCT optical biopsy is that of advancing
the technology into a needle suitable for in situ imaging in a clinical setting. To this end, we
present the design, fabrication and optical characterisation of two fibre-optic needle probes
that we use for in situ OCT needle imaging of breast tumour tissue and lymph nodes. We
demonstrate that the second of these needle probes achieves the highest sensitivity for needle
probes reported to date. Furthermore, we present the use of this needle to measure the optical
attenuation coefficients of excised, unprocessed lymph nodes in a proof of principle study.
This study demonstrates the improved capability of parametric OCT for differentiating
malignant and healthy lymph nodes. This thesis supports the possibility of OCT as an optical
biopsy tool for in situ axillary lymph node diagnosis.
8. §! L. Scolaro, B. R. Klyen, R. A. McLaughlin, P. Robbins, C. M. Saunders, S. L.
Jacques, and D. D. Sampson, "Optical property measurement with 3D-OCT to
differentiate soft tissues," presented at SPIE Photonics West BiOS, Biomedical
Applications of Light Scattering V, San Francisco, USA, 22-24 January, 2011.
9. †! R. A. McLaughlin, L. Scolaro, P. Robbins, C. Saunders, S. L. Jacques, D. D.
Sampson, “Tissue differentiation in human lymph nodes using parameterized optical
coherence tomography”, presented at SPIE Photonics West BiOS, Coherence Domain
Optical Methods and Optical Coherence Tomography in Biomedicine XIV, San
Francisco, USA, 23-30 January, 2010.
10. ! B. C. Quirk, R. A. McLaughlin, L. Scolaro, P. Robbins, C. Saunders, D. D. Sampson,
“3D-OCT imaging of ex vivo human tissue using a novel rotating needle probe”,
presented at SPIE Photonics West BiOS, Advanced Biomedical and Clinical
Diagnostic Systems VIII, San Francisco, USA, January 23-30 2010.
11. †‡ R. A. McLaughlin, L. Scolaro, P. Robbins, C. Saunders, S. L. Jacques, and D. D.
Sampson, "Mapping tissue optical attenuation to identify cancer using optical
coherence tomography," in Medical Image Computing and Computer-Assisted
Intervention – MICCAI 2009. (Eds. G. Z. Yang, D. Hawkes, D. Rueckert, A. Noble,
and C. Taylor, Springer Berlin), Lect. Notes Comput. Sci., 5762, pp. 657-664, 2009.
12. §! L. Scolaro, T. Wienhold, R. A. McLaughlin, B. R. Klyen, S. L. Jacques, and D. D.
Sampson, "Measuring tissue optical properties with needle probes using optical
coherence tomography," presented at SPIE Photonics West BiOS, Biomedical
Applications of Light Scattering III, San Jose, USA, 24-29 January, 2009.
13. †! R. A. McLaughlin, L. Scolaro, B. R. Klyen, S. Hamza, P. Robbins, C. Saunders, D.
D. Sampson, “Optical coherence tomography of human breast lymph nodes:
microstructures and metastasis”, presented at SPIE Photonics West BiOS, Coherence
xxiii
Domain Optical Methods and Optical Coherence Tomography in Biomedicine XIII,
San Jose, USA, 24-29 January 2009.
14. §! R. A. McLaughlin, L. Scolaro, B. R. Klyen, S. Hamza, P. Robbins, C. Saunders, D.
D. Sampson, “Lymph node micro-architecture can be imaged using Optical Coherence
Tomography”, presented at the 31st Annual San Antonio Breast Cancer Symposium,
San Antonio, USA, 10-14 Dec 2008, and published in Cancer Research, vol. 69, no. 2,
pp. 102S-102S, Supplement, Jan 15, 2009.
15. †! R. A. McLaughlin, L. Scolaro, B. R. Klyen, J. J. Armstrong, S. Hamza, P. Robbins,
C. Saunders, D. D. Sampson, “Can normal lymph node architecture be characterised
by optical coherence tomography?”, Proc. SPIE vol. 7139: presented at the 1st
Canterbury Workshop on Optical Coherence Tomography and Adaptive Optics,
Canterbury, UK, 8-10 September 2008, paper 19.
Other publications: 16. R. A. McLaughlin, B. C. Quirk, D. Lorenser, X. Yang, B. Y. Yeo, A. Curatolo, K. M.
Kennedy, L. Scolaro, R. W. Kirk, and D. D. Sampson, “A microscope in a needle”,
Optics and Photonics News, vol. 23, no. 12, p. 40, December 2012. (This article was
selected for the 2012 special issue “Optics in 2012”. It also featured in a video highlight
of the issue, http://www.osa-opn.org/home/articles/volume_23/december_ 2012).
xxiv
xxv
List of Acronyms ALND Axillary lymph node dissection
CNB Core-needle biopsy
CT X-ray computed tomography
DAQ Data acquisition card
FDOCT Fourier-domain OCT
FFT Fast Fourier transform
FNAC Fine needle aspiration cytology
FWHM Full width at half-maximum
GRIN Graded- or gradient-index
H&E Haematoxylin and eosin
ITC Isolated tumour cells
MEMS Micro-electro-mechanical systems
MRI Magnetic resonance imaging
NA Numerical aperture
NCF No-core fibre
NDF Neutral density filter
Near-IR Near-Infrared
OCT Optical coherence tomography
OD Optical density
PBS Phosphate-buffered saline
PET Positron emission tomography
SDOCT Spectral-domain OCT
SLNB Sentinel lymph node biopsy
SMF Single-mode fibre
SNR Signal-to-noise ratio
SSOCT Swept-source OCT
TDOCT Time-domain OCT
TIR Total internal reflection
US Ultrasound
VOA Variable optical attenuator
WD Working distance
xxvi
1
Chapter 1
Introduction 1
One pathway for the metastasis (or spread) of breast cancer to other parts of the body is
through the lymphatic system [1]. The lymphatic system transports tissue fluid, called lymph.
Its primary function is to provide an accessory route for lymph to be returned from tissues
throughout the body to the blood stream. Lymphatic organs play an important part in the
immune system: lymph nodes are the organs that facilitate the body’s immune response by
filtering the lymph as it circulates. It follows that cells of metastatic cancer usually appear first
in the lymph nodes, specifically in the lymph nodes that are closest to the site of a primary
tumour. For breast cancer, the lymph nodes most commonly involved in metastasis are those
located in the axilla (or underarm).
The presence of metastatic cancer in axillary lymph nodes is an important prognostic
indicator for patients with breast cancer [2]. Both the number of nodes involved and the
extent to which nodes are affected are important factors to determine. Nodes that contain
metastatic cancer (referred to as malignant or involved nodes) indicate the presence of a
malignant tumour and prognosis for patients with axillary metastasis is much worse than for
those without metastasis. In evaluating lymph node involvement, it has become standard
practice that only the sentinel (or first) lymph node on the lymphatic pathway that drains the
tumour is assessed. This procedure is called sentinel lymph node biopsy (SLNB) [3]. If the
sentinel node is not involved, prognosis is good and the remaining nodes are not removed. If it
is determined that one or more axillary lymph nodes are involved (contain metastatic cancer),
it is usually recommended that at least two thirds of all axillary nodes are removed as part of
the staging and treatment process [3, 4]. This procedure, called axillary lymph node dissection,
is generally performed at the same time that the primary tumour is excised.
Currently, the accepted gold-standard method to determine lymph node involvement is
excision and postoperative diagnosis via histopathology. However, removal of lymphoid tissue
often results in negative side effects, the most severe of which is swelling of the arm, called
lymphoedema [5]. As excision is currently required to make a diagnosis, patients may suffer
side effects unnecessarily if the excised nodes are later found to be negative for the presence of
2 Chapter 1 Introduction
cancer. In addition, since biopsy is usually performed while the patient is undergoing primary
tumour excision, the time available for making a diagnosis is very limited.
Clearly, a reliable alternative method of biopsy that does not require lymph node removal
would be preferable to the current standard method of excisional biopsy. Such an in situ
biopsy method could improve lymph node diagnosis in several ways, in particular:
�• in situ biopsy could be used to identify healthy (negative) sentinel lymph nodes during
breast cancer surgery, thereby eliminating the need for their removal. This could reduce
the number of healthy lymph nodes excised and avoid unnecessary risk of lymphoedema
or other side effects;
�• a pre-operative implementation of in situ biopsy could establish the presence of positive
axillary nodes prior to surgery. This could improve therapeutic and surgical planning for
patients, and enable more informed decisions to be made in relation to neo-adjuvant
systemic therapy (i.e., chemotherapy); and
�• for more definitive pre-operative diagnosis, in situ assessment could be employed
alongside traditional pre-operative excisional needle biopsy techniques (such as core
needle biopsy), in order to guide accurate sampling of nodal tissue and reduce the false
negative rates of these techniques.
Some alternative clinical imaging methods for in situ lymph node assessment have already
been investigated, including: mammography, X-ray computed tomography, positron emission
tomography, magnetic resonance imaging and ultrasonography [6]. However, these methods
have not yet proven to be sufficiently accurate or suitable for routine use. Optical diagnostic
techniques, on the other hand, may have greater potential for detecting the signature of
cancer, due to their high resolution, high detection sensitivity, low toxicity and potential for
real time assessment.
Light has long been recognised as an important tool in medicine. One of the earliest Nobel
prizes was awarded to Niels Ryberg Finsen in 1903 for his use of light therapy for the
treatment of skin disease. In modern medicine, the use of light is still important for
therapeutic applications in dermatology; however, light is more commonly used in medicine
for diagnosis. For example, optical microscopy is the fundamental tool for diagnosis of disease
in excised tissue, as it enables histological evaluation of tissue pathology. For in situ optical
assessment of tissue, the term “optical biopsy” has come into common use. “Optical biopsy”
refers to an optical measurement tool (most commonly, but not limited to, a type of
spectroscopy) that is used to diagnose tissue pathology in real-time. Although the term
Chapter 1 Introduction 3
‘biopsy’ actually refers to the removal of tissue, optical biopsy is commonly understood to
represent biopsy that does not require tissue removal. For example, diffuse optical tomography
and optical spectroscopy have been shown to reveal pathological changes in the structure and
function of breast tissue in vivo [7, 8]. The properties of light – its intensity, wavelength,
phase/group delay, degree of coherence (spatial and temporal) and polarisation – are
fundamentally affected by the micro- and macro-composition of the tissue through which it
propagates. Understanding and detecting the changes in these properties is the aim of optical
biopsy. Although in situ optical methods of tissue diagnosis are relatively new, and have not
yet been implemented clinically, they are gaining significant attention in medical research.
Optical coherence tomography (OCT) is one such modality that is being explored as a
tool for in situ optical biopsy [9]. First developed in the early 1990’s, OCT is a high-
resolution, three-dimensional optical imaging modality that measures the intensity of light
elastically backscattered from a tissue, analogous to the way ultrasound imaging measures the
reflection of sound [9, 10]. As the speed of light is much faster than that of sound, OCT
requires a more complex method of detection. Therefore, unlike ultrasound imaging, OCT
uses interferometry and a low-temporal-coherence light source to pinpoint the depth-location
of reflected light. It does so through the generation of a coherence gate, which strongly
discriminates detection of singly scattered from diffusely scattered light. The main advantage
of imaging with light is that it enables a higher resolution than imaging with sound: on the
scale of 10 – 20 μm for OCT compared to typically >100 μm for ultrasound imaging.
The resolution of OCT fills the gap between traditional microscopic techniques (such as
confocal microscopy) and clinical imaging techniques (such as magnetic resonance imaging,
computed tomography, and ultrasound) (see Figure 2-7 in Section 2.3, ‘Resolution and
imaging depth of OCT compared with other imaging modalities’). OCT resolution is lower
than that of confocal microscopy, but this is compensated for by its increased penetration
depth, which can be up to ten times greater, at around 2 to 3 mm compared to 0.2 mm. This is
possible due to the high detection sensitivity of OCT as a result of the amplification obtained
when multiplying the weak sample signal with a strong reference signal. In addition, the axial
resolution of OCT is determined by the coherence gate and is therefore decoupled from the
lateral resolution. This permits the use of a low NA objective making axial translation of the
objective beam within the sample unnecessary. Although the penetration depth of OCT does
not reach that of clinical imaging modalities (such as MRI or ultrasound), its non-invasive,
high-resolution, three-dimensional imaging capability, as well as its low cost and high
portability, make it a very attractive tool for clinical tissue assessment.
4 Chapter 1 Introduction
It is also possible to achieve ultra-high resolution with OCT, approaching 1 μm both
axially and laterally [11]. Unlike traditional microscopy, axial resolution with OCT is
decoupled from the properties of the imaging beam. This means an increase in axial resolution
does not come at the expense of a reduction in depth of field and, consequently, penetration
depth (see also Section 2.3.1, ‘Basic principles of OCT’).
One application where the advantages of OCT have been particularly beneficial is
ophthalmology. OCT is now commonly accepted as a clinical standard for visualisation of
structure and disease in the retina and for imaging of the cornea and anterior segments of the
eye [12, 13]. A literature search conducted in Elsevier’s Scopus database1 provides a useful
snapshot of the major research areas for OCT applications in biology and medicine over the
past ten years. A summary of the search results is set out in the chart in Figure 1-1 below. The
search was conducted for publications within the past ten years, with a title containing the
words “optical coherence tomography” and with words relating to specific OCT biomedical
applications, as indicated in the chart. It is evident from the graph that ophthalmology has
consistently been the most researched OCT biomedical application, with more than 1776
research publications.
Figure 1-1. Bar chart of a snapshot of results from a Scopus search of selected OCT biological applications over the past ten years. Searches were conducted on the publication title and only included English results. Search terms used were as follows: “optical coherence tomography” AND (terms in brackets under each heading) e.g. "optical coherence tomography" AND (breast OR mamm*), where the use of a * indicates a search for all words containing the truncated term.
1 http://www.scopus.com
Chapter 1 Introduction 5
In easy-to-access superficial tissues such as skin, the use of OCT has been explored for
dermatological assessment of skin cancer, scar assessment and other diseases [14, 15]. As the
resolution, penetration depth and imaging speed of OCT has improved over the past ten
years, it has steadily gained increasing attention in this field. Similarly, in dentistry, the
benefits of OCT are fast being realised, particularly because enamel is highly transparent at
near-infrared (near-IR) wavelengths [16].
OCT imaging of less accessible surfaces within the hollow organs of the body is possible
using endoscopic fibre-optic probes. In cardiology in particular, endoscopic OCT is finding a
niche for imaging atherosclerotic plaques of coronary arteries [17]. Recent growth in
commercially available intravascular OCT imaging systems has meant that cardiology has
been the second most published clinical application of OCT, as shown in Figure 1-1 above.
Recent advances in fibre optics and photonics have improved the speed and quality of
imaging and have expanded the range of applications for endoscopic OCT. Applications
where endoscopic OCT is emerging as a useful clinical imaging tool include in imaging of the
gastrointestinal tract (gastroenterology) and respiratory tract (pulmonary medicine) [18, 19].
For tissues of the gastrointestinal tract (including the oesophagus, colon and bile/pancreatic
duct), endoscopic OCT is especially useful since it enables visualisation of subsurface tissue
morphology. This is a significant advantage over conventional endoscopy, which only enables
imaging of surface features. In pulmonary medicine, endoscopic OCT has been useful for
visualising abnormal epithelia in the proximal airway and for imaging the size and shape of the
upper airway. Other applications for endoscopic OCT have included urology (bladder and
prostate), gynaecology and laryngology.
Endoscopic OCT is not limited to superficial tissues of hollow organs. The recent
development of highly miniaturised endoscopes using small fibre-optic components has
enabled OCT imaging to be translated into needles. In situ OCT imaging of solid tissues with
needle probes has shown much promise, although it has not yet been demonstrated in a
clinical study. The application of OCT needle probes has only recently begun to be explored
for imaging of tissues in the breast [20], lung [21], brain and pancreas. Imaging of tissues
within the brain for applications in neurology in animal models have particularly made
progress over the past two years [22].
One of the greatest challenges for extending clinical applications of OCT is to find
improved contrast mechanisms that can better differentiate tissue pathology. This challenge is
being addressed by the development of functional extensions of OCT, which can provide
additional information not contained in structural images. Doppler OCT is one such
6 Chapter 1 Introduction
functional extension that can detect particle motion in order to image tissue blood flow. This
enables visualisation of vasculature and blood flow velocity, which can be superimposed on
structural OCT images. Since many pathologies exhibit an abnormal vasculature, Doppler
OCT may prove useful in a number of applications including ophthalmology, dermatology,
gastroenterology and neurology [23]. Polarisation sensitive OCT (PS-OCT) is another
extension of OCT that measures the polarisation state of reflected light to detect changes in
birefringence and dichroism [24]. PS-OCT has been investigated for imaging in birefringent
tissues such as skin and muscle. Spectroscopic OCT combines spectroscopic analysis with
OCT to obtain depth-resolved tissue absorption spectra; such spectra can be used to measure
localised tissue oxygen saturation [25]. More recently, OCT elastography has been
investigated to exploit the variations in elasticity of healthy tissue and cancer in the breast
[26]. Other means of contrast, including the use of molecular probes, spectral properties and
optical properties, may also lend further capacity to OCT imaging for some applications [27].
1.1 Thesis motivation
The potential for OCT to perform optical biopsy is, therefore, well recognised in many
biological and medical applications. The use of OCT imaging for optical biopsy of cancer is a
promising, albeit challenging, area of research and there is an evident lack of exploration of
OCT for imaging of malignant lymph nodes. This thesis explores the potential of OCT in the
optical biopsy of malignant axillary lymph nodes.
A baseline study of OCT imaging of lymph nodes will be presented, along with two novel
developments that advance the capability of OCT imaging towards in situ optical diagnosis of
lymph node status. The first development uses an alternative means of OCT contrast for
differentiating healthy and malignant tissue, based on their differing optical properties. The
second explores the translation of OCT imaging hardware into miniature optical probes
inserted through hollow needles.
1.2 Structure of this thesis
Chapter 2 - Background
Chapter 2 provides the background on the role of lymph nodes in the spread of breast cancer.
The standard method for assessing the involvement of lymph nodes based on excision and
histological evaluation is described, as well as the possible side effects of lymphoid tissue
Chapter 1 Introduction 7
excision. A summary of alternative techniques is then presented, followed by an introduction
to OCT. The specifications of the OCT systems used in the investigations of this thesis are
also outlined.
Chapter 3 – OCT imaging of axillary lymph nodes
This chapter investigates the potential use of OCT (in its standard form) for imaging of
excised human axillary lymph nodes. To this end, the tissue composition of healthy and
malignant axillary lymph nodes is discussed, with reference to representative histology images
of healthy and malignant lymph nodes. A short summary of the current literature on OCT
imaging of lymph nodes is also set out. The following section, reproduced from [28], presents
an investigation into the ability of OCT to image lymph node micro-architecture and identify
metastasis, based on comparison with corresponding histology images. The chapter concludes
with a discussion of some of the shortcomings of intensity-based image contrast for
differentiating malignant and healthy lymph nodes, and the need for improved contrast.
Chapter 4 – OCT for optical property measurement in lymph nodes
In Chapter 4, we introduce and demonstrate a method for improving the contrast of OCT
imaging, based on extracting a tissue optical property, specifically the attenuation coefficient.
We visualise extracted attenuation coefficients in the form of en face parametric maps
(parametric images). There are various methods for modelling and analysing OCT image
intensity to extract the attenuation coefficient, and a literature review of current methods is
discussed. Our chosen method is subsequently described, based on the single-scattering model
of OCT image intensity. The next part of this chapter presents an amalgamation of two
previously published studies ([29] and [30]), which report on the use of the single-scattering
model to extract the relative and quantitative attenuation coefficients of healthy and
abnormal (including malignant) lymph nodes. These results support the use of parametric
imaging to aid interpretation of the OCT appearance of lymph nodes and facilitate their
diagnosis.
Chapter 5 – Needle-based OCT
The fifth chapter presents a development of the OCT imaging technology to achieve imaging
of tissue in situ, thereby enabling tissue analysis to be conducted without the need for excision.
This chapter describes the design and implementation of two novel OCT needle probes – the
second being an advanced version of the first. The material for these sections is reproduced
from our previous publications: [31] and [32]. The application of each needle is demonstrated
8 Chapter 1 Introduction
for imaging of excised lymph nodes and breast tumour tissue. Finally, we present the capability
of the second needle to measure optical properties of tissue. This work provides a proof of
principle demonstration that needle OCT parametric imaging can be used to differentiate
healthy and malignant lymph nodes.
Chapter 6 - Conclusion
The final chapter summarises our main findings, which are as follows: firstly, that OCT can
identify the micro-architectural appearance of healthy lymph nodes and that cancer has an
identifiable appearance in OCT images characterised by high scattering intensity; secondly,
that parametric imaging of the local optical attenuation coefficient, extracted from OCT
image data, can improve contrast and interpretation of OCT images and guide tissue
segmentation; and finally, that OCT parametric imaging can successfully be implemented in
optical needles and be used to differentiate human axillary lymph nodes and breast tumour
tissue. Based on these conclusions, we discuss the potential of OCT to achieve in situ lymph
node biopsy and discuss some alternative applications for this work that may ultimately prove
useful. We finish with a discussion of future work that would be required to realise our aim of
enabling practical and reliable in situ optical biopsy of axillary lymph nodes.
9
Chapter 2
Background 2
In this chapter we give the background of lymph node involvement in breast cancer and
explain the motivation for using optical coherence tomography (OCT) to assess the status of
axillary lymph nodes. We first discuss the importance of lymph node assessment for cancer
staging and the current method to diagnose cancer in excised lymph nodes by excision and
histopathological evaluation. We outline the side effects of lymph node removal including
lymphoedema and give a brief review of the medical imaging literature focusing on alternative
methods for lymph node assessment. Optical imaging by OCT is discussed as an alternative to
achieve lymph node ‘optical biopsy’. We discuss the advantages and disadvantages of OCT as
an optical biopsy tool and present the principles of OCT and a summary of the OCT systems
used in the investigations reported in this thesis.
2.1 Breast cancer and lymph node metastasis
Breast cancer is the most common form of cancer in women, accounting for almost one in
four female cancer diagnoses worldwide in 2008 [33]. Most breast cancers have their origins at
the terminal ductal lobular unit of the breast with cellular proliferation occurring within these
units and within the adjoining ductal structures. For early stage cancer, the tumour is usually
confined to these ducts and is referred to as ductal carcinoma in situ. Breast cancers progress
from in situ to invasive when malignant cells infiltrate the duct walls and invade the
surrounding breast tissue. Invasive breast cancer becomes metastatic when malignant cells
break away from the main tumour and spread to other organs of the body through either the
blood stream or the lymphatic system. The risk of malignant cells metastasising through the
lymphatic system is directly proportional to the size of the primary breast tumour [2]. Invasive
breast cancer is the second leading cause of cancer-related female deaths in Australia and the
US [34]. Accurate determination of the extent and spread of disease (or staging) is a critical
component of the care and management of individuals with breast cancer.
10 Chapter 2 Background
2.1.1 Disease progression
The lymphatic system provides one pathway for the dissemination of malignant cells in the
body, and lymph nodes represent a common site of metastatic involvement. Like the
circulatory system, the lymphatic system is made up of vessels that transport fluid. In the case
of the lymphatic system this fluid is lymph or tissue fluid that drains from nearby tissues.
Deposits of metastatic cells are, therefore, usually observed first in the regional lymph
nodes. Lymph nodes are a major component of the lymphatic system that lie in clusters or
chains along the lymphatic vessels at strategic locations where they drain nearby anatomic
regions, such as the neck, abdomen and axilla. They facilitate the body’s immune response
through mechanical filtration of the foreign bodies in the lymph and through recognition and
processing of antigens. The regional lymph nodes of the breast are those clustered in the axilla
as illustrated in Figure 2-1. There are about 20-35 lymph nodes in the region of the axilla and
this is usually the first anatomic site that is involved in breast cancer metastasis.
Figure 2-1. Illustration of breast cancer and its metastasis to lymph nodes. (a) Metastatic breast tumour spreads to the loco-regional lymph nodes in the axilla. The sentinel node is the first along the lymphatic pathway draining the tumour and is usually the first site where metastatic cancer is found. (b) Cancer metastasises to other organs of the body through the blood stream or through the lymphatic system. From: The website of the National Cancer Institute, “Breast Cancer Treatment”, 2012. [Online]. Available: http://www.cancer.gov/cancertopics/pdq/treatment/breast/Patient. [Accessed: 11 December 2012]. Used with permission from Terese Winslow, 2012.
2.1 Breast cancer and lymph node metastasis 11
2.1.2 Staging and treatment
Involvement of axillary lymph nodes in cancer metastasis provides the single most important
indicator for prognosis in early breast cancer [35]. Involved lymph nodes represent an invasive
tumour biology and, therefore, an increased risk of secondary cancer and mortality. The stage
of breast disease is classified according to tumour size and the degree of lymph node
involvement, with both the number and location of histologically positive lymph nodes
having prognostic significance. For staging of breast cancer: Stage 0 represents no nodal
involvement, Stages I and II may have some nodal involvement and Stages IIIA, IIIC and IV
always have nodal involvement [36]. The stage is indicative of the severity of disease with later
stages being more severe.
Where lymph nodes are involved, the standard recommendation for therapy is often
complete axillary clearance; that is, removal of all accessible axillary lymph nodes in order to
eliminate the nodes that may contain cancer and prevent further cancer spread.
2.1.3 Morphology of lymph nodes
Lymph nodes are small and roughly bean-shaped organs, varying in size from 2 mm to 20 mm.
They lie within a supporting framework of adipose tissue that surrounds the node and usually
occupies the concave space at the central hilum where blood and lymphatic vessels enter and
exit. A diagram of the typical morphology of a lymph node is shown in Figure 2-2.
Lymph nodes have a complex architecture in which a variety of cell populations are
arranged in distinct interfacing compartments. The entire node is enclosed by a fibrous
capsule of dense connective tissue with trabeculae that extend deep into the parenchyma as
part of the supporting stromal framework. The lymph node parenchyma consists mainly of
small immune cells called lymphocytes in various stages of maturation, as well as supporting
cells, reticular fibres (made up of fine collagen fibrils) and blood vessels. Lymph nodes can be
roughly divided into the following compartmental regions or zones as indicated in Figure 2-2:
cortex (light blue in Figure 2-2), in which primary and secondary follicles are generally
located; paracortex (light green), situated just below the cortex and between cortical follicles;
and medulla, which lies deeper in the node and consists of medullary cords (dark green) and
medullary sinuses (pale yellow). The medullary sinuses eventually join to form the efferent
lymphatic vessels.
12 Chapter 2 Background
Figure 2-2. Basic structure of a typical lymph node. The lymph node is roughly bean shaped with an indentation at the hilum where blood and lymphatic vessels exit. It is cushioned by surrounding adipose tissue (not shown). The internal structures include a capsule of fibrous connective tissue (brown) and a parenchyma of lymphocytes and supporting cells compartmentalised into cortex (light purple) with cortical follicles (dark purple), paracortex(light green) and medullary cords (dark green) and sinuses. Adapted from: Stevens A, Lowe J. Histology. London:Gower Medical Publications, 1992, pp. 90-95.
Lymphatic flow into the lymph node occurs through afferent lymphatic channels that
enter at the capsule and drain through subcapsular sinuses into the medullary sinuses.
Outward flow of lymph occurs through the efferent lymphatic channels at the hilum. It is
often the case, therefore, that early metastatic deposits are identified first in the subcapsular
sinuses [37]. However, metastatic cells can also spread through the whole node to completely
replace normal lymph node tissue. Malignant nodes containing metastatic tumour are
generally larger and harder than healthy benign reactive lymph nodes. More detail on the
cellular morphology of healthy and malignant lymph nodes is given in Chapter 3, in Section
3.2.2 and Section 3.3.3, respectively.
2.1.4 Standard histopathological evaluation
The current gold standard method for identification of metastasis in lymph nodes is
histopathological evaluation. To facilitate accurate assessment, excised lymph nodes are first
bisected into 2 mm portions along the long axis. Each portion is then evaluated by
histopathology. Standard histological preparation involves fixation, dehydration, mounting
and thin slice sectioning of the tissue, followed by appropriate staining to visualise the
morphology of cells at high resolution using an optical microscope. The most widely used
stain in pathological diagnosis, and the one utilised for the work in this thesis, is haematoxylin
and eosin (H&E) that stains basophilic structures (such as nuclei) blue, and eosinophilic
structures (such as cell cytoplasm and collagen) pink.
2.1 Breast cancer and lymph node metastasis 13
2.1.5 Side-effects of histopathological evaluation
The standard method for lymph node diagnosis is histopathological evaluation. However,
histopathology requires nodes to be excised before they can be assessed. In addition to causing
trauma for the patient, there are significant side effects associated with lymph node removal,
including chronic lymphoedema, restricted arm movement, partial sensory loss (due to
sensory nerve injury) and pain. Lymphoedema is a swelling of the arm, breast or chest wall that
is caused by an accumulation of lymph in the tissue. An example of the effect of lymphoedema
on a patient after lymph node removal is shown in Figure 2-3. Medical management of the
condition can include intermittent pneumatic compression, decongestive physiotherapy and
elevation. These options usually require lifetime behaviour modification. In a survey of the
literature, Petrek et al. found that patients undergoing axillary lymph node dissection
(ALND) developed lymphoedema in 6% to 36% of cases, depending on such factors as weight
Historically, nodal involvement has been determined by ALND, in which at least two thirds
of lymph nodes in the axilla are removed and evaluated by histopathology. Although ALND
remains the "gold standard" for sensitivity and accuracy of detection, it carries a higher
morbidity than the alternative and now widely accepted technique of sentinel lymph node
biopsy (SLNB) [38, 39]. SLNB requires the removal of only the sentinel lymph node, which is
the first draining lymph node on the direct lymphatic pathway from a primary tumour,
although there may be more than one sentinel node. If the sentinel node does not contain
metastasis, ALND can be avoided with minimal increased risk of mortality [39]. Thus, SLNB
is an alternative to ALND that significantly reduces morbidity associated with lymphoedema.
14 Chapter 2 Background
The location of the sentinel lymph node is determined by ‘mapping’ through the injection
of a radioactive colloid and/or a blue dye (patent blue) near the site of the tumour [3]. The
radioactive colloid is injected several hours prior to surgery and a lymphoscintigram image (an
example of which is shown in Figure 2-4) roughly locates the sentinel lymph node. During the
surgery, a hand-held gamma probe is used to accurately locate the ‘hot’ node. Patent blue dye
is a complementary method that is used to visually identify the node by its blue appearance
after injecting the dye near the tumour at the commencement of surgery.
Figure 2-4: Lymphoscintigram transmission images taken prior to surgery. Anterior (A) and right lateral (B) images taken 30 minutes after injection of the radionuclide dye in the left breast (arrow). Focal uptake can be observed in the sentinel lymph node (dashed arrow). Originally published in JNM: Pandit-Taskar, N. et al., “Organ and fetal absorbed dose estimates from 99mTc-sulfur colloid lymphoscintigraphy and sentinel node localisation in breast cancer patients”, J. Nucl. Med., 2006, vol. 47:1202-1208, by the society of Nuclear Medicine and Molecular Imaging Inc.
Diagnosis of sentinel node status is usually performed while the patient is undergoing
surgery, so that the surgeon can proceed with a full therapeutic axillary clearance if the node is
found to be positive for metastasis. Therefore, to minimise surgery time and risk to the
patient, fast assessment of the node is required and this is usually accomplished using frozen
section histology or imprint cytology. These techniques typically require 15-30 minutes to
complete. For frozen section histology, a section is made by snap freezing the tissue to enable
slicing, followed by staining and viewing of the section under a microscope. Alternatively,
imprint cytology takes a cell ‘smear’, rather than a section, from an internal cut surface of the
node that is then microscopically examined. Although these techniques are fast to perform,
their sensitivity and specificity (53 – 75% and 98% – 100%, respectively [40, 41]) are much
lower than for standard histopathological assessment because of the error involved in bisecting
the node at the location of a metastatic deposit. In particular, micro-metastases, which are
defined as metastases less than 0.2 mm in size are often missed. Detection of micro-metastases
is a significant challenge for lymph node diagnosis, but there is also considerable debate as to
whether they are significant for prognosis [42]. In accordance with low sensitivity, there is a
recognised false negative rate of diagnosis with these techniques: in a survey of the literature
2.2 Alternative techniques for in situ assessment of lymph node status 15
Kim et al. found that an average of 7.3% of breast cancer patients with positive axillary nodes
were falsely diagnosed as negative at biopsy [43]. Thus, excised sentinel nodes are usually also
submitted for detailed histopathological analysis after surgery. Additional therapeutic ALND
(that is axillary clearance) may then be required in a second surgery based on the results of
histopathology; however, this occupies additional valuable surgery time and can cause
increased trauma for the patient.
The risk of lymphoedema after SLNB is lower than for ALND; however, there is still an
associated risk. In one study the occurrence of lymphoedema for SLNB patients was around
6.9% [44]. In the same study, 3.8% of patients who underwent SLNB reported restricted
movement and 8.6% reported feeling numbness at 6 months after surgery. Furthermore, only
42% of sentinel nodes contain metastases [43], indicating that uninvolved healthy lymph
nodes are excised in a significant proportion of cases.
2.1.7 Motivation for a new assessment technique
The motivation for an in situ intra-operative assessment technique is to avoid the unnecessary
excision of lymph nodes when the axilla is free of malignancy. A high accuracy in situ method
for assessment of node status prior to excision could reduce the rate of removal of negative
sentinel or axillary nodes and improve patient outcome and well being. This method could be
employed intra-operatively or pre-operatively. The advantage of an intra-operative assessment
technique is that it would work well with current SLNB procedures and, if fast enough could
also reduce the time required for lymph node assessment during surgery.
2.2 Alternative techniques for in situ assessment of lymph node status
SLNB is now commonly employed for breast lymph node assessment worldwide. However,
the search for a less invasive clinical method for axillary lymph node assessment is of increasing
interest and has led to research growth in a number of current and new modalities for lymph
node assessment. Alternative modalities for lymph node assessment may be broadly classified
into pre-operative needle biopsy techniques, standard clinical imaging techniques and novel
optical imaging techniques. The success of these modalities to assess the status of axillary
lymph nodes depends on their ability to identify either the molecular, morphological or
vasculature signature of metastasis with minimal side effects. Thus, we discuss the most
16 Chapter 2 Background
significant modalities here, in light of their diagnostic accuracy (sensitivity and specificity) for
detecting lymph node involvement.
2.2.1 Pre-operative needle biopsy
Pre-operative needle biopsy uses a hollow-core needle to obtain a sample of cells for fine-
needle aspiration cytology (FNAC) or a section of tissue for core-needle biopsy (CNB) [45-
48]. Extracted samples are examined by histopathological evaluation, similar to SLNB. For
FNAC, an aspirate of cells is taken from the node using a small diameter (21 – 23-gauge, 0.8 –
0.65 mm outer diameter) needle. The aspirate is smeared onto a microscope slide and the cell
morphology is assessed for cancer. Alternatively, CNB uses a larger needle (14-gauge, 2.1 mm
outer diameter) with an outer cutting cannula that is used to excise a small section of tissue
after insertion. The specificity for predicting positive nodes by either FNAC or CNB is very
high, close to 100%, since cancer cells are easy to identify under histopathology. On the other
hand, it is difficult to locate the nodes within the body using these techniques and there is
significant error in sampling diagnostically significant tissue, that is, tissue from the node as
opposed to external adipose or connective tissue. Thus, sensitivity can be very low ranging
from 57% to 87% for FNAC [45, 49] and from 40% to 94% for CNB [50, 51]. For CNB,
there is also the added complication that the procedure is invasive and problems can arise
from damaging major vessels and nerves [51]. Some work has been done to improve sampling
through better guidance. Both of these techniques must be performed under image guidance
using an external imaging modality to direct the needle to the location of the node. This is
usually accomplished by ultrasound imaging. Recent studies have looked into better mapping
of lymph nodes in order to improve guidance either using colour-Doppler ultrasonography or
photo-acoustic tomography with contrast agents [50].
2.2.2 Pre-operative imaging
Pre-operative imaging techniques may provide significant benefits since these have the
potential for non-invasive assessment of the axilla. The most notable current in vivo imaging
techniques being investigated for this purpose are: mammography, X-ray computed
tomography (CT), positron emission tomography (PET), ultrasound (US) and magnetic
resonance imaging (MRI) [6].
Mammography is the standard screening tool for initial detection of tumours of the breast.
A positive diagnosis is based on increased absorption of X-rays from high-density tumour
tissue and the presence of micro-calcifications. Mammographic imaging extended to the axilla
2.2 Alternative techniques for in situ assessment of lymph node status 17
detects pathological lymph nodes based rather on their size. However, contrast based on size is
not considered an accurate means of differentiating healthy and malignant lymph nodes and,
thus, the accuracy of mammography is very low [6, 52].
The extension to three-dimensional imaging with CT, as opposed to two-dimensional
mammography, has the advantage that more information may be obtained over a large field of
view and from different imaging planes. Also, imaging is more comfortable for the patient,
since it does not require tissue compression and can be captured with the patient lying down.
However, initial results for CT imaging of lymph nodes in the axilla showed 50% to 82%
sensitivity and 75% to 94% specificity for detection of nodal involvement [53, 54]. This high
variability in accuracy can be attributed to the criterion for differentiation being again only
size [55].
Most recent studies show that PET is more sensitive in imaging the axilla for nodal disease,
with one study reporting a sensitivity of 90% overall, and up to 100% for patients in the T1
(largest) tumour size category [56]. Contrast is based on uptake of radiolabelled glucose
analogues, since there is a higher molecular uptake in fast metabolising cancer cells. However,
the specificity of PET is highly variable (and can be <50%), especially for micro-metastatic
deposits in small lymph nodes, and so the potential for staging the axilla with this modality is
limited [56, 57].
Ultrasound imaging, on the other hand, has demonstrated greater potential for pre-
operative imaging of axillary nodes, especially when combined with FNAC [58, 59] for which
it is becoming an important adjunct clinical imaging modality. In a summary of previous
studies of ultrasound imaging of the axilla, Alvarez et al. found that the sensitivity for
detecting pathological nodes with axillary sonography varied from 44% to 90% depending on
several factors including palpability of the nodes and the number of criterion used. Specificity
varied from 75% to 100% when implemented with FNAC [59]. Standard US assessment is
based on shape and internal morphological appearance, rather than size, to predict lymph
node status. Healthy nodes typically have a thin hypoechogenic cortex and a central echogenic
hilum with a thin border between (Figure 2-5). The length to width ratio (l/w) is an indicator
of node shape and is usually >2 for healthy nodes. A pathological node, on the other hand,
usually exhibits a rounded appearance due to eccentric thickening of the cortex, and l/w is
usually less than 1.5. Absence of the hilum is also highly suggestive of metastasis. Recent
enhancements in US imaging that use colour Doppler flow to image vasculature as an
additional contrast mechanism, have increased the sensitivity and specificity of conventional
[69-71]; fluorescence microscopy, including confocal and multi-photon microscopy [72-76];
and, more recently, optical coherence tomography (OCT) [9, 10, 77, 78], which forms the
focus of the investigations in this thesis and which we discuss further in the following section.
Of those techniques mentioned above, optical spectroscopy has demonstrated the greatest
potential for clinical use. In a review by Bigio and Mourant [79], studies of elastic-scattering
spectroscopy were identified as potential tools for in vivo detection of cancer in the bladder
and gastrointestinal tract, with sensitivities ranging from 60% to 100% and specificities from
80% to 97%. Preliminary research of elastic-scattering spectroscopy for breast cancer detection
and probe guidance during brain surgery were also highlighted. So too were clinical studies of
fluorescence spectroscopy for diagnosis of disease in the oesophagus, cervix and colon, and
preliminary studies of Raman spectroscopy for accurate optical biopsy of cancer in skin, breast
and cervical tissue. The sensitivity and specificity of fluorescence spectroscopy were found
over several studies to range from 83% to 100% and 73% to 98%, respectively.
In summary, optical diagnostic techniques offer the possibilities of higher resolution,
higher sensitivity and faster detection than current clinical imaging modalities. These benefits
could also provide improved disease management by enabling earlier diagnosis and immediate
treatment. The recognised success of these techniques in a range of clinical research areas has
helped drive their progression towards potential clinical applications over the past decade in
particular.
2.3 Background of optical coherence tomography
OCT is a high-resolution, three-dimensional optical imaging modality that has shown great
potential for disease diagnosis in a range of applications. OCT measures elastically
20 Chapter 2 Background
backscattered light to form structural images of tissue, analogous to how ultrasound imaging
measures the reflection of sound.
Since the velocity of light is much faster than that of sound, its reflection cannot be
measured by direct detection based on time-of-flight. Rather, OCT uses interferometric
detection to indirectly measure the location of the reflection by correlating two optical waves,
one reflected from the tissue sample and one from a reference mirror.
The general configuration of an OCT system is that of a Michelson interferometer, which
consists of a broadband light source, a coupler leading to sample and reference arms and a
detector (see Figure 2-9). Depth scans of sample reflectivity are built up either by rapidly
changing the length of the reference path or by performing an inverse Fourier-transform on
the optical spectrum of the interfered waves (these two methods of depth-scanning are
discussed in Section 2.3.1 with further detail given in references [10, 77] and [80]). Cross-
sectional and three-dimensional (3-D) images of tissue are then synthesised from laterally
adjacent depth scans.
The source wavelengths used in OCT are in the near-IR range, in the so-called ‘diagnostic
window’, where absorption due to tissue chromophores is low. In this wavelength range,
scattering dominates over absorption allowing light to penetrate sufficiently to form an image,
usually to a maximum depth of a few millimetres. Figure 2-6 shows some typical components
of tissue that contribute to absorption[81-83]. In the figure, the absorbance of water has been
multiplied by 500 to better visualise it within the plot. Note the high absorbance of water on
the long-wavelength side, and of melanin and deoxyhaemoglobin on the short-wavelength side
of the diagnostic window.
Figure 2-6. Diagram of the optical ‘diagnostic window’ showing the spectral absorption for a range of tissue chromophores. The diagnostic window exists in the near-infrared wavelengths (800–1300 nm), where light attenuation/extinction due to absorption is relatively low. The absorbance of water has been multiplied by 500 to enable its visualisation on the scale of the plot. The arrows indicate typical source wavelengths used for OCT. Composite figure adapted from data in [81] [82] [83].
2.3 Background of optical coherence tomography 21
Light penetration is determined by scattering and absorption. Scattering occurs as a result of
the heterogeneous components that make up tissues including cells, organelles and fibres of
collagen. In general, scattering decreases monotonically with increasing wavelength [84]. The
most common wavelengths for OCT are therefore 1300 nm and 840 nm: although deeper
penetration is achieved at 1300 nm, axial resolution is better at 840 nm (see Equation (2.1)).
The axial resolution of OCT is generally higher than that of standard clinical imaging
systems and approaches that of confocal microscopy as depicted in Figure 2-7. Compared to
conventional microscopy, a major advantage of OCT is its sensitivity, which can reach the
physical limit set by photon shot noise; typically -90 to -110 dB below a perfect reflector. Such
high sensitivity enables penetration up to 2 mm in some tissues, compared to several hundred
microns at best for confocal microscopy. High resolution is also maintained over this depth,
since, unlike confocal microscopy, axial resolution in OCT is decoupled from the numerical
aperture (NA) of the optical system, so that accessing greater imaging depths (through the use
of a low NA objective) does not come at the expense of limited axial resolution [77] (also
illustrated in Figure 2-8). Typical OCT axial resolutions are on the order of 5–10 µm, but
ultrahigh resolutions around 1-2 µm have also been achieved [85]. The low penetration depth
of OCT compared to current medical imaging modalities is a limitation of the technology,
but this may be overcome by the use of compact endoscopic probes and imaging needles that
can be directed to the site of interest. Such probes are now realisable with the development of
new laser and fibre-optic technologies. Advances in detector and acquisition technology are
Figure 2-7. Axial resolution and imaging depth of OCT compared with other imaging modalities. OCT imaging resolution and penetration depth sit at the boundary between potential in situ medical imaging and ex vivo microscopic imaging modalities. The ‘pendulum’ length represents imaging depth, and the ‘sphere’ size represents resolution. From: T. Hillman, “Comparison of OCT with other imaging modalities”, 2009. [Online]. Available: http://obel.ee.uwa.edu.au/research /oct/intro. [Accessed 20 December 2012].
22 Chapter 2 Background
also enabling video-rate, 3-D OCT imaging. Of particular interest for clinical use is functional
OCT imaging, including polarisation-sensitive OCT, Doppler OCT, OCT elastography and
wavelength-dependent OCT. These extensions possess enormous potential for improving
OCT contrast for differentiation of tissue pathology. Overall, OCT is an attractive option for
optical biopsy because it permits real-time, in situ imaging of tissue microstructure with high-
resolution, enabling morphological differences between normal and neoplastic tissues to be
visualised without the need for excision or the addition of exogenous contrast agents. It is also
highly portable, relatively easy to implement and low-cost, making it highly practical for intra-
operative use.
2.3.1 Basic principles of OCT
OCT depth scanning can be achieved using one of two techniques: time-domain low-
coherence interferometry or spectral interferometry. The first implementations of OCT were
based on time-domain OCT (TDOCT) and employed optical low-coherence reflectometry
to generate depth- or A-scans [9, 86]. A standard TDOCT system uses a low temporal
coherence (broadband) light source and a splitter that sends the light into two paths – one
terminating at the sample and the other at a scanning reference mirror as shown in Figure 2-9.
Interference fringes occur only if the (round-trip) optical path lengths of the reference and
sample beams coincide within the coherence length of the source. Therefore, a depth scan is
acquired by scanning the reference mirror and recording the envelope of the generated
interference waveform.
Frequency-domain OCT (FDOCT) is a more recent technique that employs spectral
interferometric techniques to achieve depth sectioning with much faster speeds than
TDOCT [10, 80, 87]. High acquisition speed is possible because the depth scan is generated
by performing an inverse fast Fourier transform (FFT) on the cross spectral density obtained
by interference of the reference and sample beams. Therefore the entire depth scan is obtained
in one computation avoiding the need for the slow scanning reference mirror. The other
major advantage of FDOCT is that it is possible to achieve superior sensitivity with this
technique. This topic has been treated extensively in the literature [80, 88]. Significantly,
unlike TDOCT, the signal-to-noise ratio (SNR) of spectral techniques is independent of the
source optical bandwidth.
There are two distinct methods that have been employed to obtain the spectral signal in
FDOCT. The first uses rapid wavelength tuning of a narrowband source to acquire time-
encoded wavelength information and is called swept-source OCT (SSOCT). The second
2.3 Background of optical coherence tomography 23
method uses a spectrometer to analyse the space-encoded spectral content of the signal and is
called spectral radar or spectral-domain OCT (SDOCT).
For both TDOCT and FDOCT, axial resolution z is given by the full width at half-
maximum (FWHM) of the OCT axial response. Assuming a Gaussian source with a FWHM
bandwidth of , and mean wavelength the axial resolution is [77],
z = 2ln22
,
(2.1)
Transverse resolution is decoupled from axial resolution and is entirely dependent on the
sample arm optics as demonstrated in Figure 2-8. It is given by the FWHM diameter of the
intensity distribution of the sample beam at its waist, with W 0 the 1/e2 diameter of the
sample beam at its waist. For a beam with a 1/e2 waist diameter of W 0 ' incident upon a lens
with focal length f the transverse resolution x, is given by [89],
x = 2 ln2W0 =2 ln2 f
W0 '. (2.2)
The Rayleigh range zR, describes the useful depth-of-field of imaging, where the width of the
beam is within a factor zR times its minimum value. It is given by,
2zR =2 W0
2
. (2.3)
Therefore, a focused beam with a low numerical aperture (NA ≈ / ( W 0) [89]) is usually
desired to achieve reasonable resolution over a large depth-of-field.
Figure 2-8. The OCT sample optics determines transverse resolution and axial depth of field. The coherence gate determines axial resolution. Adapted from: J. G. Fujimoto, “Optical coherence tomography: Introduction”, in Handbook of Optical Coherence Tomography, B. E. Bouma, G. Tearney, eds. New York: Marcel Dekker, Inc., 2001, pp. 1-40.
The final parameter of OCT image quality is sensitivity. In the shot-noise limit, the
sensitivity is proportional to the optical power incident on the sample and inversely
proportional to the A-scan rate and can be calculated by [77],
24 Chapter 2 Background
S =
hc1( )PS PR
PS + PR( )B, (2.4)
where is the quantum efficiency of the detector, h is Plank’s constant, c is the speed of
light, is the intensity splitting ratio of the interferometer splitter (assuming no loss), PS
and PR are the optical powers incident on the sample and reference mirror, respectively and
B is the electronic detection bandwidth, which is a function of the depth scan velocity in
TDOCT, and of the acquisition rate in FDOCT (see references [10] and [90] for more detail
on bandwidth and sensitivity in OCT).
The investigations in this thesis are based on results from three OCT systems. These are:
an in-house TDOCT benchtop system operating at 1310 nm; a fast-scanning commercial
SSOCT system [91] (Thorlabs, USA) operating at 1325 nm; and a second SSOCT system at
1310 nm built during this thesis as a portable system optimised for OCT needle imaging. The
source centre wavelengths for these systems are all around 1300 nm so that they will achieve
high penetration depth when used to image whole lymph nodes. A description of these
systems is given in the following section.
2.3.2 OCT systems used in this thesis
TDOCT
Figure 2-9 shows a schematic of the fibre-based TDOCT benchtop system used for the first
investigation of this thesis (Chapter 3). The splitter/coupler sends polarised light from the
superluminescent diode light source (DenseLight SLED DL-BD9-CS31159A) into the
reference and sample arms. The reference arm is terminated by a frequency-domain optical
delay line (FD-ODL) that rapidly scans the axial path length of the reference arm to achieve
depth sectioning at the sample. It also enables compensation of the dispersion mismatch
between the two arms [92]. Light in the sample arm is directed onto the sample by a weakly
focusing (0.07 NA) triplet lens. The beam is scanned laterally across the lens using a
galvanometer-scanning mirror placed at the back-focus of the lens to accomplish planar
imaging at the sample. Scanning in the third dimension is achieved by translating the sample
on a translation stage.
A circulator at the source arm of the coupler enables balanced detection to reduce
intensity noise. This is achieved by a dual-balanced photodetector (bandwidth = 100 Hz to
3 MHz), which rejects the common-mode DC component of the outputs from the circulator
and the coupler. The dual-balanced photo-detected difference signal is bandpass filtered and
2.3 Background of optical coherence tomography 25
logarithmically demodulated in hardware before analog-to-digital conversion and acquisition
(PCI-6111E, National Instruments, Austin, Texas). Cross-sectional images from this system
were displayed in real-time in Labview using a custom user interface. This system has also been
described previously [93, 94]. A summary of its specifications is given in Table 2-1.
Sensitivity ( S ) -112 dB (shot noise limited) -108 dB
A-scan rate 500 Hz 500 Hz
Depth scan range in air ( Z ) 2.26 mm 2.3 mm
Transverse scan range* ( X , Y )
— 3.7 mm × 6.0 mm
Typical sampling density (pixel size) ( Z , X ,Y ) — 2.1 µm × 3.4 µm × 5.9 µm
Power at sample ( PS ) — 2.9 mW *Determined directly by the settings of the galvanometer scanner and translation stage at the sample
26 Chapter 2 Background
The advantages of TDOCT are its relative low complexity and ease of implementation,
with components for this system being readily attainable (the broadband source and
associated passive components of the interferometer were upgraded during this thesis). Its
main disadvantage is its very slow imaging rate of 2 seconds per B-scan with 3-D scans taking
up to 35 minutes, which is not feasible for clinical studies.
SSOCT
Schematics of the Thorlabs 1325 nm SSOCT system (OCS1300SS) and the custom portable
1310 nm SSOCT system used in this thesis are shown in Figure 2-10 and Figure 2-11,
respectively. Both systems use a frequency clock (Mach-Zehnder interferometer (MZI) clock)
so that sampling at the interferometer output is linear in k-space (wavenumber), since the
Fourier transform pair is wavenumber and group delay. Both systems employ balanced
detection, although this is slightly different in the custom-built system, which was made for
maximum flexibility for using OCT needle probes in both dual-arm Michelson configuration
and common path Fizeau configuration. Thus, the balancing arm in this system is provided by
the output of a second splitter at the source, which takes power from the source to balance its
intensity noise as shown in Figure 2-11.
Scanning at the sample arm of the Thorlabs system was performed by an XY galvo-
canning mirror pair enclosed in a detachable handheld scan head, which included a visible
CCD camera to view the scan area. The mirrors translate the beam over a maximum area of
10 × 10 mm at the sample. All data acquisition and processing was performed via the
integrated software package and PC provided by Thorlabs. The system achieved high imaging
rates; however, the axial and lateral resolutions were lower than the TDOCT system. Also it
Figure 2-10. Schematic diagram of the Thorlabs SSOCT system: SS, swept laser source; FC, fibre coupler; PC, polarisation controller; CIR, circulator; C, collimator; AP, adjustable pinhole variable attenuator; M, mirror; BD, balanced detector; DAQ, data acquisition card; SD, XY scanners driver; CCD, camera; OBJ, objective. From Thorlabs OCS1300SS Swept Source OCT System User Guide.
2.3 Background of optical coherence tomography 27
Table 2-2. Theoretical and measured performance parameters for the Thorlabs SSOCT system.
Parameter Provided by supplier Measured
Centre wavelength ( 0 ) 1325 nm 1325 nm
Source bandwidth ( B3dB ) 100 nm 100 nm
Average source power ( P0 ) 10 mW 13.6 mW
Axial resolution in air ( zFWHM, intensity )
12 μm 17 µm
Transverse resolution ( xFWHM ) 15 μm 16 µm
Rayleigh range ( zR ) 430 μm 416 µm
Sensitivity ( S ) -100 dB (shot noise limited) -91 dB
A-scan rate (sweep rate) 16 kHz 16 kHz
Depth scan range in air ( Z ) 3.0 mm 2.8 mm
Transverse scan range ( X , Y ) 10 mm × 10 mm (max) 4 mm × 4 mm (Typical)
Sampling density (pixel size) ( Z , X ,Y )
— 5.3μm × 5μm × 5μm (Typical)
Depth scan range in air ( Z ) 5.0 mW 4.7 mW
was difficult to adjust and optimise the processing and image acquisition parameters for our
application. This also meant that replacing the sample arm with a needle sample arm would
not be possible: for example, the scanning control was integrated into the manufacturers
software assuming XY galvo scanning, but this could not be altered to allow needle scanning
control. For these reasons we built the second SSOCT scanner shown in Figure 2-11, which
was also more compact and portable to be used for OCT needle imaging at the hospital.
Further details of this system along with the needle sample arm scanning are given in Chapter
5. A summary of the specifications for each system is given in Table 2-2 and Table 2-3.
The major advantage of FDOCT over TDOCT is that imaging is much faster (Thorlabs
system scan rate = 25 Hz B-scan rate). Also, FDOCT is known to have a sensitivity advantage
of up to 20 dB over TDOCT [80]. The reference arm is also stationary and more compact
than TDOCT making it easy to align and portable. In the case of SSOCT, both reference arm
alignment and signal detection is simple and easily made portable. One disadvantage of
SSOCT is that, due to limitations in the bandwidth of currently available swept sources, the
axial resolution is limited; although advances in source technology, such as Fourier domain
mode-locked wavelength swept lasers and polygon scanners are leading to improvements in
source bandwidth and speed [95]. Also dispersion compensation is slightly more difficult and
28 Chapter 2 Background
Figure 2-11. Schematic diagram of the custom-built portable SSOCT system: SS, swept source; VOA, variable optical attenuator; DAQ, data acquisition card; PD, photodetector; tr, sweep trigger; clk, k-linear source clock.
Table 2-3. Theoretical and measured performance parameters for the 1310 nm custom-built needle SSOCT system.
Parameter Theoretical Measured
Centre wavelength ( 0 ) 1310 nm 1310 nm
Source bandwidth ( B3dB ) 100 nm 100 nm
Average source power ( P0 ) 18 mW (min) 41.5 mW
Axial resolution in air ( zFWHM, intensity ) 13 μm 13 μm
Transverse resolution ( xFWHM )
— *
Rayleigh range ( zR ) — *
Sensitivity ( S ) -117 dB (shot noise limited) -114 dB
A-scan rate (sweep rate) 50 kHz 50 kHz
Depth scan range in air ( Z ) 5 mm 4.7 mm
Transverse scan range ( X , Y ) — 360° × 6 mm (Typical)
Sampling density (pixel size) ( Z , X ,Y ) — 4.9μm × 0.5° × 5μm (Typical)
Depth scan range in air ( Z ) — 18 mW *The transverse resolution and Rayleigh range are determined by the optics of the needle probe used in each experiment (see Chapter 5 for resolutions of the needle probes used in this thesis )
generally needs to be implemented numerically in software after acquisition. Possible
improvements to the OCT technology, as well as the use of functional and multimodal
extension to OCT imaging, are briefly discussed in Chapter 6. The current work focuses on
the use of standard OCT and in the next chapter we present the application of TDOCT for
imaging of excised healthy and malignant human axillary lymph nodes.