Get Complete Project Material File(s) Now! »
Temporal low-coherence and slice sectioning in FFOCT
Axial scanning is performed via a motorized translation stage allowing sequential image acquisition with slices as thin as 1 micron. The setup sectioning ability is directly related to the low temporal coherence of the light source. Interferences occur only when the light from both arms has travelled a nearly identical “optical distance” as shown on figure II.2.
The low temporal coherence gate therefore allows only interferences within half of the coherence length. Consequently, only a slight difference in pathlength travelled causes the interferometric image to be blurred by background noise. When light reflected by the reference mirror interferes with the light reflected or backscattered by the sample, micro-structures contained within the volume are filtered out from the specimen. This virtual zone is a slice orthogonal to the objective axis, located at a depth inside the object defined by an optical path length difference of zero. If the object arm is moved forward or backward, a different sectioning volume is imaged. The sectioning thickness is determined by the width of the fringe envelope amplitude. This is commonly measured by the full-width at half maximum (FWHM) of the signal amplitude and is equivalent to the axial resolution as detailed further in this chapter. The tomographic image is finally obtained by combining interferometric images with a phase shift accomplished by displacement of the reference mirror with a piezoelectric translation stage.
Image acquisition method
The full-field “en-face” tomographic image is obtained via a phase-shift method in order to extract the coherent interferometric signal from the specimen. The backscattered light intensity received by each pixel (x,y) of the CCD camera over time can be expressed by the following equation (interference signal of two waves): I ( x , y , t)= I 0 4 [Rinc(x , y)Rref (x , y)+ 2√ Robj(x , y )×Rref (x , y)cos(Φ( x , y )+ Ψ)]( 4) where I0 is the photon flux at the entrance of the interferometer, Rinc(x, y) is the proportion of light reflected by the object that does not interfere, due to other reflectors in the sample arm path combined with external signal noise from the instrument; Rref represents the homogeneous reference mirror reflectivity, since the system is illuminated via a Khöler illuminator providing a quasi uniform beam of light. Within the interference term, Robj(x, y) represents the proportion of light reflected by the object that interferes with the reference mirror and more precisely, Robj(x, y) is the reflectivity distribution from structures contained in the sample’s coherence volume and thus the “en face” tomographic image that actually needs to be extracted. As the result of the interference pattern, the cosine term is composed of the unknown phase difference, Φ between the reference mirror and the object signal; and the additional term Ψ accounts for the phase-shift induced by the displacement of the reference mirror.
A phase-shift method is used to reduce the amount of incoherent signal and to extract the amplitude of the interference (the fringe envelope) from the sample (i.e. Robj(x, y)). Practically, only two to four images are sufficient to reduce the incoherent signal. For example, by acquiring four successive values of Ψ : 0, π/2, π, 3π/2, and by computing the squared difference of shifted frame, it becomes possible to isolate the term: √ Robj (x , y )×Rref ( x , y) , proportional to Robj (x , y)×Rref (x , y ) or √ Robj (x , y ) , the amplitude of the backscattered signal intensity.
Tomography and coherence plane dynamic adjustment
A 3-D tomographic image is acquired by displacement of the focal plane at different depths below the surface and a stack of “en-face” images is thus collected. As the objective is displaced along the z-axis, the focus is shifted forward whereas the coherence plane is moved backward as shown on figure II.4. This characteristic can benefit refractive index measurements –and has been used for that purpose in highly scattering tissues . For the purpose of imaging deep within a specimen, a dynamic adjustment is necessary to compensate for the loss in resolution in relation to depth. In this setup, it is done by aligning the coherence plane with the objective focal plane during the scanning process. The following compensation factor is applied to the reference arm as detailed in  : Full-Field OCT system: design principles and performance δ(z )=2z (n ‘ ²−n²) n (5) where δ(z ) is the optical path difference or phase shift between the object and the reference arm, a difference induced by the mismatch of refractive index between the biological sample n’, and the immersion medium n (water or silicone oil) and dependent upon the depth being probed. The plane of zero path difference or coherence plane is therefore aligned with the focal plane by displacing the reference arm -the objective and reference mirror are then kept in focus. The cut-off depth at which the adjustment is necessary is highly dependent upon the biological sample being imaged. This limit can be approximated by comparing the ratio δ(z )/2n ‘ relative to the depth of field. For the microscope objectives used during this work, mainly 10X 0.3 NA water immersion, the depth of field
is typically around 8.5 μm (wavelength centred at 700 nm and Δλ=125 nm). According to the refractive index of the specimen studied, the following table provide a summary of the threshold depth at which defocus correction is deemed necessary.
System performance: spatial resolution
As previously seen, the transverse and axial resolution in FFOCT are uncorrelated. The sectioning ability along the z-axis is solely dependent upon the light source temporal coherence: a low coherence or a large spectrum allows a highly resolved sectioning. In contrast to the spectrum of ultra-short femtosecond lasers, a thermal light source provides a particularly smooth and stable spectrum. It avoids emission lines spikes, bumps or ripples that could potentially cause side lobes in the coherence function and creates artefacts.
Nonetheless, the effective spectrum is actually limited by the spectral response of the camera (CCD or CMOS5). With a CMOS image sensor camera (Photonfocus®), the effective spectrum is centred at l=700 nm and provides a bandwidth of Dl=125 nm at full width half maximum (FWHM), following the usual formula (valid for a Gaussian spectral profile): d z=2ln2 nπ ( λ2 Δ λ ) (6).
Thus, the theoretical axial resolution achieved in a medium is below 1 μm with (dz = 0.8 μm) with a refractive index n= 1.377 for a typical biological sample. Experimental results show rather an axial resolution in the range of 1.1–1.4 μm depending on the camera and setup configuration. This loss in resolution is certainly due to dispersion occurring in one of both arms; a slight mismatch from the ideal zero path difference can also easily deteriorate the point spread function along the zaxis. As detailed in the previous section, several compensation methods can be implemented.
Since water is the main constituent of biological tissues, the use of water-immersion microscope objectives help to minimize dispersion mismatch. In addition, for the experimental setup, thin glass plates, placed in both arms of the interferometer, can be tilted to compensate for residual dispersion.
Large field imaging of ex-vivo breast tissues.
To date, only a few studies have been reported about the capability of high-resolution OCT to aid in the visualization of normal and pathologic breast structures at the micron scale . An isotropic resolution up to 3 μm is commonly referred to high-resolution or ultra high-resolution in optical coherence tomography literature[2–4]. As a comparison, a similar level of resolution is provided by low or medium power microscopy (4X magnification). According to histopathologists consulted, a majority of common pathologies could potentially be detected and assessed with low power magnification on stained histological slides. A morphological assessment could therefore in theory be sufficient for most pathologies. This work attempts to address this hypothesis by methodically comparing tomographic information with gold standard histology.
Despite advances in early detection of breast cancers through breast imaging techniques such as ultrasound, mammography or CT scans, breast cancer remains among women the most prevalent cancer worldwide . In addition, the increase of more sensitive screening techniques for breast cancer has lead to a rise in surgical procedures. However, 14–40% of patients still require a second surgical procedure due to positive or close margins , . Positive margins are defined as tumour cells appearing directly at the cut edge of the excised specimen. Residual tumour left bear the risk of local recurrence and require additional surgery for the patient [8-11].
Among the intra-operative margin assessment techniques, frozen section analysis has been shown to reduce the rate of second operations to about 20% , however its rationale in clinical routine is being questioned in the literature , . Although the accuracy of frozensection is very high , [15–18], it has been suggested inappropriate with very small tumours (<1 cm)  and may also compromise final diagnosis by leaving insufficient material for permanent paraffin embedding or for tumour markers in molecular analysis.
A second issue with frozen-section analysis is the cost benefit rationale. Initially, it was used during surgery with a diagnostic purpose and dictated further tumour removal if repetitive positive margins were found during surgery. Cost-benefit was evident as immediate diagnosis could avoid a potential second operation for the patient. As E.Singletary pointed out: “in theory, a patient scheduled for lumpectomy who is found to have repeatedly positive margins after multiple reexcisions would be a candidate for an immediate mastectomy6. However, many institutions are reluctant to perform the mastectomy in the same surgery under these circumstances” .
The main reasons are the psychological effects for the patient of not expecting major surgery such as a mastectomy and the lack of preparation regarding a breast-conserving surgery. Patients scheduled for a mastectomy are now widely offered immediate reconstruction. Likewise such procedure also requires extensive time and coordination among specialists and cannot be immediately scheduled without a minimum notice. Therefore, the cost-benefit of frozen-section with
treatment intend has altered.
Core-needle biopsy study
Needle biopsy of breast masses as alternative to open surgical biopsy is a widely performed procedure. It is recognized as highly accurate and cost-saving when tissue sampling is possible. Core-needle biopsy (CNB) or fine-needle biopsy (FNAB) are the two main methods.
Difference between FNAB and CNB is needle’s diameter (<0.8mm vs 2.1mm respectively). Usually if the lump is palpable, a fine-needle biopsy is sufficient in most cases. Otherwise, if there is suspicion of a solid or cloudy fluid, a core-needle biopsy is preferred under image-guidance using either stereotactic mammography or ultrasound.
Table of contents :
Chapter I. Principles and context of OCT imaging in biological tissues
I.1 Conventional OCT and biomedical imaging context
I.1.1 OCT versus other imaging modalities
I.1.2 Principles of traditional OCT
Fourier or Frequency-domain OCT
Comparative advantages of OCT systems
I.1.3 Performances of conventional OCT
I.1.4 Selected applications
I.2 Optical properties of biological tissues
I.2.1 Light – tissue interaction
Absorption, therapeutic window and wavelength dependence
Scattering mechanisms in tissues
Geometrical optics approximation:
The Mie solution to Maxwell’s equations:
The Rayleigh solution to Maxwell’s equations:
Chapter II. Full-Field OCT system: design principles and performance
II.1 Full-field OCT: basic principles
II.1.1 Description of the FF-OCT setup
II.1.2 Temporal low-coherence and slice sectioning in FFOCT
II.1.3 Image acquisition method
II.1.4 Tomography and coherence plane dynamic adjustment
II.1.5 Image post-processing
II.2 System performance: spatial resolution
II.3 Compact clinical setup
Chapter III. Breast ex-vivo imaging: from laboratory to clinical setting
III.1 Large field imaging of ex-vivo breast tissues
III.1.2 Material and method
Study design and imaging protocol
Specimen selection and preparation
III.2 Core-needle biopsy study
III.2.2 Method and protocol
Other organ: kidney lesion
III.3 Discussion and Conclusion
Chapter IV. Contrast enhancement strategies: assessment and validation.
IV.1 Assessment of mapping optical attenuation coefficients in breast tissues
IV.1.1 Context and background
IV.1.2 Material and methods
Specimen selection and instrument
Scattering coefficient analysis
Image acquisition and processing
IV.2 Feasibility of an experimental setup to measure the static elastic properties of a breast tissue
IV.2.1 Background and context
IV.2.2 Experimental setup cross-correlation-based method
Chapter V. Advances in biology: in-vivo imaging of Drosophila melanogaster
Drosophila melanogaster model
V.2 Material and method
Major steps of the pupal phase
Adult fruit fly
Chapter VI. Infrared Full-Field OCT and penetration depth improvement
VI.1 InGaAs FFOCT setup
VI.1.1 Background and objectives
VI.1.2 Material and method
Si camera system
VI.1.3 Performance comparison
VI.2 Penetration depth assessment in biological tissues
VI.2.2 Material and method
OCT signal attenuation and penetration depth
Annexe: Résumé en Français
Contexte et problématique
Principes et État de l’art en OCT dans les milieux biologiques
Dispositifs d’OCT « plein-champ »
Etudes cliniques sur lésions mammaires et biopsies
Pistes d’amélioration du contraste endogène
Avancées en biologie : imagerie in-vivo du cycle d’une métamorphose.