Open Access
Add to BibSonomyAbstract
We present a Bessel beam illumination FDOCT setup using a FDML Swept Source at 1300nm with up to 440kHz A-scan rate, and discuss its advantages for structural and functional imaging of highly scattering samples. An extended focus is achieved due to the Bessel beam that preserves its lateral extend over a large depth range. Furthermore, Bessel beams exhibit a self-reconstruction property that allows imaging even behind obstacles such as hairs on skin. Decoupling the illumination from the Gaussian detection increases the global sensitivity and enables dark field imaging. Dark field imaging is useful to avoid strong reflexes from the sample surface that adversely affect the sensitivity due to the limited dynamic range of high speed 8bit acquisition cards. In addition the possibility of contrasting capillaries with high sensitivity is shown, using inter-B-scan speckle variance analysis. We demonstrate intrinsic advantages of the extended focus configuration, in particular the reduction of the phase decorrelation effect below vessels leading to improved axial vessel definition.
©2011 Optical Society of America
1. Introduction
Optical Coherence Tomography (OCT) is a noninvasive three-dimensional imaging modality using using low coherence interferometry [1]. Thanks to the achievable high acquisition speed due to recent developments in light source and detector technology, OCT has demonstrated its particular effectiveness for in-vivo imaging and in particular for biomedical applications. In contrast to other optical imaging methods, the depth resolution is decoupled from the lateral one, the latter being still determined by the focusing objective and its numerical aperture (NA). Usually low numerical apertures are used in OCT to keep a long depth of focus (DOF). However, this comes at the price of low lateral resolution according to Gauss optics and it becomes clear that a trade-off appears: between the lateral resolution, which is related to the beam diameter, and the focal range, usually given by twice the Rayleigh range. If high lateral resolution is required, dynamic focusing [2], where coherence gate and confocal gate are kept at the same depth position during axial scanning, can be employed. A natural development in this direction is en-face OCT [3], where the depth scan is performed slowly, whereas the lateral scanning speed may profit from resonant scanners. However, for volumetric in-vivo measurements, the relative long measurement time required by these techniques has to be reduced to keep longitudinal motion artifacts small [4]. In order to maintain high sensitivity with ultrahigh-speed and ultra-high resolution, Fourier domain optical coherence tomography (FDOCT), where a depth profile is acquired in parallel, is the best candidate [5–9]. Benefiting from this advantage in combination with high lateral resolution requires however methods to extend the DOF beyond the limit given by Gaussian optics.
Different approaches in FDOCT have been taken to overcome this trade-off: they involve either C-mode scanning, by stitching in-focus volumes together [10], numerical algorithms involving the complex diffraction field [11–14] that are limited by low signal-to-noise-ratio (SNR) of out-of-focus signals, or direct modifications of the optical setup like a multiple beam approach, which however increases the system complexity [15,16]. A first attempt in the replacement of the Gaussian illumination beam, for time domain OCT, has been using a conical lens or axicon in front of the sample instead of an objective [17]. This configuration is in principle well suited for FDOCT with its parallel depth recording [18] since a collimated beam propagating through an axicon gives rise to a field with a prominent central lobe having constant transverse extension along the optical axis. The intensity distribution is described by a Bessel function [19]. The disadvantages of this configuration for in-vivo imaging are first the missing possibility to steer the beam, and second, a strong loss in sensitivity as compared to confocal OCT system due to double pass through the axicon. A different approach to produce a Bessel field illumination is to generate a ring shaped light distribution in the back-focal plane of the sample objective. This is achieved via ring shaped slit. The most energy efficient way is to illuminate the ring mask via axicon lens, as has been shown in [20]. The use of the axicon leaves the ring mask in principle obsolete, however improved performance in particular concerning intrinsic dark field property can be achieved [21]. The sensitivity critical double pass through the mask is overcome by decoupling the detection and the illumination path [20]. The detection is confocal and exhibits higher detection sensitivity. Compared to standard OCT configurations, such extended focus scheme offers an increased focal range while preserving enough sensitivity for in-vivo imaging.
Functional extension, like for example polarization sensitive or Doppler OCT [1], has raised growing interest because of relevant additional tissue contrast and complementary physiologic information not available from standard backscattering OCT tomograms. Especially microcirculation imaging with OCT has gained from recent technological developments for high-speed FDOCT [22,23]. Imaging microvasculature has in particular shown great potential for cancer research [24], but could also help for treatment monitoring, e.g. in photodynamic therapy (PDT) [25]. During the last years, a number of strategies for contrasting microvasculature based on FDOCT have been introduced. Contrast between flow and static tissue can be obtained by determining phase or amplitude changes either between A-scans [26–30] or successive tomograms [23,24,31], or from spatial frequency analysis across a full tomogram [32–34]. Speckle variance analysis [25] is especially suitable for swept source based system, because it is independent from any trigger jitter. Also the fact that it does not require bulk motion correction alleviates the post-processing effort as compared to phase difference or phase variance techniques.
The present work demonstrates for the first time the advantages of extended focus OCT (xf-OCT) for in-vivo structural, as well as functional skin imaging. The xf-OCT system is equipped with a high-speed Fourier domain mode locking (FDML) swept source operating at 1300nm center wavelength [36–38] with a maximum A-scan rate of 440kHz. Choosing the center wavelength at 1300nm is especially apt for imaging of highly scattering tissue due to the reduced scattering at longer wavelengths [35]. In particular, we will discuss the advantages of intrinsic features of the extended focus scheme for structural and functional imaging: its dark field property [21] as well as the self-reconstruction property of Bessel beams [39], a novel aspect for OCT imaging.
2. Experimental
2.1 Optical setup
The xf-OCT imaging system (Fig. 1 with box a) is a swept source based system equipped with a FDML laser centered at 1310nm with 140nm full-bandwidth, giving an axial resolution of 12µm in air. The sensitivity roll-off was measured as −1.2dB/mm. The cavity filter is driven at 55kHz. Only one sweep direction was used for the measurement, however the output power is spread over both sweep directions leading to a SNR equivalent to 110kHz. The light from the source is directed to a fiber-based Mach-Zehnder type interferometer with a first 80/20 fiber coupler to maximize the power on the sample. 20% of the light is directed in the reference arm for which the polarization as well as the dispersion can be controlled and adjusted to optimize the interference signal. In the sample arm, the light is collimated and passes a conical lens or axicon (A) (apex angle of 5°, fused silica, DelMar optics). The axicon is followed by a 4f system (L1-L2) that produces the ring shape intensity distribution in its Fourier plane allowing for a proper spatial separation between illumination and detection. A small mirror (M, 4mm of diameter for a ring diameter of 6mm) placed at this position decouples the detection (blue path) from the illumination to avoid a double pass through the axicon and associated strong loss in sensitivity [20]. It is an intensity efficient solution for the detection of backscattered sample light as compared to the use of a beam-splitter, and furthermore, it allows filtering of low spatial frequency components, e.g. stray light from the tip of the axicon in the illumination direction. The lens (L3) after the 4f system focuses the ring pattern on the mirror of the fast axis scanning system. The scanner is designed to produce a virtual single pivot-point scan for large beam diameter [Leica Patent US2006/0176536 A1]. A second 4f relay system (L4,L5) is used to image the ring into the back-focal plane of the objective (L6), giving rise to the Bessel beam field pattern at the sample.
Fig. 1 Optical Setup with a and b for the extended focus and standard configuration respectively. Blue: detection path. SS: Swept source, FC: Fiber coupler, PC: Polarization control, DC: Dispersion compensation, A: Axicon, M: Mirror, L1 to L6: Lenses, Galvo: Scanning mirrors, DBD: Dual-balanced detector, Circ: Circulator.
The system was optically designed to produce a Bessel beam with a central lobe size (1/e2) of 5.5µm over an axial distance of 1.6mm. The ring center diameter at the principal plane of the objective is 8mm, giving an effective NA of 0.13.
The detection is of standard Gaussian or confocal type with the detection fiber serving as pinhole. The detection optics was designed for a spot size at the sample plane of 15µm (that can be compared to the 5.5µm of the illumination) and a Rayleigh range of about 135µm. The detection NA of 0.056 is smaller than that of the illumination and thus produces a dark field effect by rejecting directly reflected light. Furthermore, reflected light from a slightly tilted plane surface at the sample position has a low coupling efficiency into the detection fiber since it will be collimated at the detection fiber tip. A second detection configuration, called later extended tight focus, was used in order to show the high lateral resolution property of the Bessel beam illumination. A spot size at the sample plane of 9µm (NA 0.09) allows a tighter apodization of the Bessel field and thus a stronger attenuation of the side lobes. However, the DOF is limited by the smaller Rayleigh range.
The sample and reference lights are then directed to a 50/50 fiber coupler. The interference signal is measured by a dual balanced detector (350 Mhz, PDB130C, Thorlabs). The signal was digitized at 250MS/s using a data acquisition board (DAQ) with a 8bit analog-to-digital converter (ADC) (1GS/s, ATS9870, Alazartech). A low-pass filter (90MHz cut-off frequency, Mini-circuits) was used to prevent aliasing and to reduce the electronic noise. The acquisition and the processing were performed with Labview 2010 (National Instruments). The FDML laser offers high stable operation over a measurement duration and permits thus to avoid the use of a reference interference signal on the second channel of the DAQ or on a k-clock. The resampling to correct for the source’s intrinsic k-nonlinearity was performed numerically by using a reference signal acquired with a mirror at the sample position prior to measurement. The high sensitivity of OCT permits to get sufficient reference signal from a tilted mirror even with the dark field configuration.
The above extended focus setup can easily be changed to a standard OCT configuration as shown in Fig. 1 (box b) by using the detection path of the previous setup. The light from the source passes a circulator and is split by a 50/50 fiber coupler into the reference arm of the previous system, and the sample arm, which is now connected to the previous detection path. The focal length of the collimator was increased to produce a theoretical spot size of 6µm (NA of 0.137), a value close to the theoretical transverse resolution of the Bessel beam system, with a corresponding 20µm Rayleigh range. In that case, the diameter of the small mirror M has to be adjusted to account for the larger beam diameter. The standard setup, i.e. with Gaussian illumination and detection, was used for performance comparison.
2.2 Performance of standard OCT vs. xf-OCT
Figure 2 compares the cross sections of the cylindrically symmetric theoretical lateral point spread function (PSF) for different depths calculated for the central wavelength for: the standard confocal, the extended tight focus and the extended focus configurations (see section 2.1). It demonstrates the better lateral resolution available close to the focus in the extended tight focus case and the larger axial extend of the central lobe in the extended focus case. It also shows the expected higher SNR of the standard configuration at the focal position, which soon becomes smaller than both extended focus configurations with increasing distance from the focal point.
Fig. 2 Theoretical lateral PSFs for the central wavelength in dB comparing the different configurations. The side lobes exhibit a stronger attenuation close to the focus in the extended tight focus, leading to a better resolution at that position, while the central lobe is less attenuated along the depth in the extended focus leading to a larger axial depth.
The resolution of the systems was experimentally compared by using a 4.5µm polystyrene microsphere on a glass plate that was measured at different successive depths. The size of this sphere produces a signal closely related to the PSF of the system. Figure 3(a) , 3(b) and 3(c) shows the signal for the standard, extended tight focus and extended focus configuration respectively. The central lobe of the Bessel distribution is well conserved over a large depth range. The number of visible side lobes increases with distance from the focal position, while in the standard case, the PSF rapidly enlarges. Note, that the PSF for the standard configuration could only be measured for a smaller depth range than for the extended focus cases. A quantitative evaluation can be performed by plotting the average radial profile represented in Fig. 3(d) for selected depths. Close to the focus position, the three configurations exhibit almost the same full width half maximum (FWHM). The FWHM for the extended configurations remains constant along the 1.6mm axial distance while a clear increase in the standard spot size can be seen. As expected, the extended tight focus shows a better attenuation of the side lobes close to the focal position. The reduced contrast of the side lobes can be explained by the finite size of the microsphere that don’t act as point scatterers. A common way to describe the lateral resolution of a system is to use the FWHM of the PSF. As visible in Fig. 3, the extended configuration exhibits better resolution performance than the standard case over the full depth range. This is essentially due to the fact that the side lobes of the extended focus PSF are much smaller in amplitude than the central lobe.
Fig. 3 Measurement with a 4.5µm microsphere on a glass plate for several depths acquired with standard, extended tight focus and extended focus configuration in a, b and c respectively (numbers in µm; linear intensity scale). d: Average radial intensity profiles for selected depths showing similar FWHM at the focal position and the increase of the spot size for standard configuration with depth.
3. Results
The palm of the hand of healthy volunteers was measured in vivo. The palm was in contact with a glass plate for stability. The measurements presented were taken with the extended and the standard system in the same skin area, thus exhibiting the same biological features.
3.1 Structural imaging
The 3D volume data set consists of 800 B-Scans each composed of 800 A-Scans scanned over 2x2mm. The power on the sample was 7.5 and 5.7 mW for the extended focus setup and the standard one respectively.
The tomograms were converted to a logarithmic [dB] scale. The same scale with respect to minimum and maximum intensity values was used to quantitatively compare the signal evolution along depth, as presented in Fig. 5 . In order to objectively compare image quality the tomograms were adjusted to express same brightness at the noise floor and maximum signal value. The en-face pictures were adjusted to have the same mean value (at the middle of the dynamic range) and standard deviation of their bell-shaped histogram intensity distributions.
3.1.1 Comparison standard to extended focus (xf) configuration
Figure 4 shows tomograms (a, b & c) and the average of 6 tomograms (d,e & f) taken from the 3D volume and acquired with standard, the extended tight focus and the extended focus configuration, on the left, in the middle and on the right respectively. The axial delay was converted to geometric path length by using the group refractive index of water of 1.34. Obviously, the extended focus offers deeper tissue imaging as compared to the standard and also to the extended tight focus configuration. The axial imaging range seems now to be limited by the scattering properties of the sample itself; dark low scattering structures, that relate to subcutaneous fat, are predominant at this depth and location.
Fig. 4 Comparison of tomograms. a, b & c: Single tomogram, d, e & f: Average of 6 tomograms taken from the 3D data set, a & d: Standard, b & e: Extended tight focus, c & f: Extended focus. Scale bar denotes 250µm in every picture. Lateral marks indicate focal depth positions.
The sensitivity of the standard configuration was measured as 107dB using the conventional measurement procedure with a mirror and neutral density filter at the sample position. The sensitivity of the extended configurations cannot be measured this way, because of the intrinsic dark field illumination. However it is possible to compare systems based on the measured SNR of a common scattering sample. A quantitative appreciation of the extended focus advantages can be seen in Fig. 5 where the mean intensity value, calculated over the tomogram width, is plotted along the geometric depth. The curves were vertically shifted to get the same signal from the transition stratum corneum/stratum lucidum, while preserving the original dynamic range. The focus being set at this position for each system permits a direct evaluation of the SNR along the depth. The smaller noise floor of the standard configuration compared to the extended focus expresses its higher SNR. The difference of 5 to 7dB corresponds to what was previously measured [20]. An equivalent sensitivity of 100dB for the extended configuration can thus be postulated. Obviously, the better SNR for the standard configuration is confined to the confocal range. Beyond this range the Bessel illumination exhibits improved SNR resulting in larger depth of field.
Figure 6 shows 3D renderings of a 2x2mm section acquired with the extended and extended tight focus configuration on the left and right respectively. The volume on the right was cut to present surfaces close to the epidermis/dermis transition to appreciate the lateral resolution offered by the system.
Fig. 6 3D renderings of 2x2mm sections acquired with a: Extended focus configuration and different depths positions of Fig. 7 highlighted in blue, b: Extended tight focus configuration and cut to appreciate the lateral resolution of the system.
In order to compare the lateral resolution performance for the different configurations, en-face pictures shown in Fig. 7 for the depths indicated in Fig. 6 (left) are presented. They represent the standard (left column), tight focus xf- (middle column), and xf-configuration (right column). Small structural features are resolvable with high contrast with the extended tight focus configuration, while the extended focus allows seeing structure beyond the depth range of the other techniques. The transverse resolution of structures at larger depths appears degraded because of the increased detection spot size at that position and also because of multiple scattering.
Fig. 7 Comparison of en-face pictures for different depths. Left: Standard, middle: Extended tight focus, right: Extended focus. a, b & c: 240µm, d, e & f: 300µm, g, h & i: 380µm, j, k & l: 910µm deep under the surface respectively. Scale bar denotes 250µm in every picture.
3.1.2 Self-reconstruction property of Bessel beam illumination
Thanks to its particular conical illumination scheme, the extended focus system suffers less from shadowing artifacts due to surface features that exhibit strong attenuation. To illustrate this phenomenon, measurements were taken with the same previously described parameters but on the dorsal part of the index finger at positions where hairs are present (Fig. 8 ). The pictures are not post-processed. In the standard configuration, black stripes are visible below the hair core and smaller ones below the hair edges. In the xf-OCT configuration, less attenuation under the hair is visible and en-face views are almost free of shadowing artifacts below a defined depth. It can be explained by the concentric ring illumination: a small particle on the surface will not disturb the part of the beam that illuminates a deeper part of the sample. The effect depends on the hair size and its relative position to the focus. Since the detection is still performed axially, the suppression of shadowing artifacts is not complete and depends on the opacity of the obstacle.
Fig. 8 Effect of a hair on the signal attenuation along the depth. a, c & e: Standard, b, d & f: xf-configuration. a & b: Tomograms, c, d, e & f: En-face views at 2 different depths indicated by the white lines in the left tomogram showing a progressing reduction of the hair’s influence while it remains constant with the standard configuration. Scale bar denote 250µm in every picture.
3.1.3 Dark field property
To appreciate the dark field property of the xf-OCT configuration, measurements were performed while keeping the same glass plate orientation (almost perpendicular to the optical axis) for the extended focus and the standard setup. In the standard case, it is not possible to use the full power without getting saturation artifacts. The sensitivity has to be adjusted so that the reflection from the glass plate fits into the dynamic range of the acquisition board. Practically, the power of the reference arm was adjusted so that the signal of the glass plate’s reflection is close to saturation. The resulting sensitivity is then substantially lower since less reference arm power can be applied. It is then directly possible to compare both pictures in Fig. 9 . The picture taken with the standard system exhibits a low penetration range that can be explained by the lower sensitivity. Moreover, the first strong reflection masks structural information close to the surface.
Fig. 9 Tomograms taken with the same glass plate orientation, almost perpendicular to the optical axis, with the reference power adjusted to avoid getting out of the range of the ADC. Left: Standard showing a limited penetration depth as well as a masked first interface, right: Extended focus remaining unaffected by the glass plate reflection. Scale bar denotes 250µm.
This dark field property is very useful when using 8 bit DAQ boards that are currently of special interest for high-speed applications. The limited dynamic range is not compatible with a strong reflection coming from the first interface. The attenuation offered by the extended focus illumination permits to maintain the whole dynamic range for the structural information and even employing 8 bit ADC will not lead to substantial loss of SNR [40]
The configuration is in general advantageous to keep the focal region at a constant depth within the sample, which is difficult to achieve with a tilted glass plate used for fixation. Also, samples that exhibit strong specular reflexes can be imaged with optimal SNR.
3.2 Microcirculation imaging
In order to contrast microcirculation, we chose to calculate the speckle variance [25] between successive intensity tomograms due to its particular compatibility with swept source OCT, since it is not affected by any trigger jitter. The speckle variance is a fast way to display motion occurring in the sample, as it does not require bulk motion correction, in contrast to phase difference or phase variance techniques. No registration was performed during the processing.
The slow axis scanner is driven by a multiple-step function. It permits measuring successive B-Scans at the same position, giving an almost perfect correlation for static tissue. A speckle variance tomogram is calculated between 2 successive B-Scans. The mean value of every variance tomogram is plotted against time and permits to set manually a threshold to reject pictures with large variance values resulting from strong motion artifacts. The remaining variance tomograms taken at the same position are then averaged. 10 B-Scans were taken at each position; the true sampling in the slow axis direction is then given by the total acquisition of 1000 B-Scans divided by 10 for each position. Without any rejected frame due to sample motion, one would then average a maximum of 9 variance tomograms per position.
Figure 10 shows the result of averaging variance tomograms at one position for both illumination types. The vessels are represented by white values compared to the black static tissue that exhibits small variance. The difference in the static tissue brightness between both techniques can be explained by the smaller SNR of the extended focus scheme.
Fig. 10 Result of averaged variance tomograms at a single position. The static tissue appears black while motion is represented by white values. Blood vessels are visible, however their axial extend determination is limited by decorrelation. Left: Standard, right: Extended focus. Green arrows pointing the skin surface in contact with the glass plate. Scale bar denotes 250µm.
We observe in Fig. 10 that the vessels exhibit strong tails in depth. This effect can be explained by phase decorrelation due to multiple scattering in blood as well as its inhomogeneity in refractive index. In order to investigate this phenomenon further, profiles along the depth are calculated. A total of 8 vessels are selected in each averaged variance tomogram, being located at about the same depth and having approximately the same width and variance value. The depth profile is first averaged for each vessel along its width and then smoothed in depth by a moving average (5 pixels). The slight axial offset that may remain between vessels is corrected and the average of the 8 curves is calculated and normalized to its maximum value for comparison. Finally, the axial offset between the depth profile for each technique is also corrected by matching the position of maximum signal value.
Figure 11 shows the resulting curves. The extended focus curve exhibits an axially better defined vessel size and a steeper decrease of the variance signal. This can be explained by the fact that not the whole illumination light is crossing the blood vessel, in a similar fashion as the self-reconstruction property. The photons producing the image of sub-vessel structures are crossing the blood vessel at least one time for the axial detection, while at least 2 times in a standard configuration. This effect depends of course on the vessel size and its position relative to the focus. The observation of less multiple scattering artifacts below blood vessels might be an indication of a general advantage of using Bessel beams for imaging due to its self-reconstruction property, as has been demonstrated for microscopy [39]. However, a thorough analysis goes beyond the scope of this article and is subject of ongoing investigations.
Fig. 11 Normalized mean variance depth profile through a vessel showing a better defined vessel axial size and a steeper decrease of the signal with the extended focus scheme.
The better axial definition of the vessel cross sections allows for 3D visualization without the need of complex strategies to suppress the tail below the vessels [24]. The 3D variance volume is first smoothed along the axial direction by applying a moving average over 5 pixels and then thresholded. The value of the threshold is set for every A-scan to half the sum of the mean and the maximum value. Finally, a Gaussian Blur (σ = 2 pixels) was applied to improve the vessel connectivity. Figure 12 shows a 3D rendering of the resulting volume. A large dynamic range of vessel sizes is visible in Media 1 and Media 2 over different depths. Large vessels are encountered at large depths of about 900µm while smaller ones are present at the transition epidermis/dermis at about 350µm. The thresholding/blurring as well as the 3D rendering have the consequence to lower the visibility of the smallest vessels with size below 20µm diameter.
Fig. 12 3D rendering of processed speckle variance stacks. 2x2mm section acquired with the extended focus setup. Left: Volume showing a large dynamic range of vessels for different depths (Media 1). Right: Volume taken at a different position showing a more planar vessel structure (Media 2).
Figure 13 shows an overlay of the circulation on the structural information. This visualization gives a conceivable overview of the true vessel locations within the skin layers. Figure 13(a) offers a 3D rendering that permits to appreciate the better axial localization of vessels. Figure 13(b) shows an en-face projection of a 3D volume for the smaller depth that exhibits significant variance signal. It is close to the epidermis/dermis transition and shows bright points. They are thought to correspond to almost vertical capillaries that support the dermal papilla. Figure 13(c) shows the en-face projection for depths where planar vessel networks are visible. It is linked at some positions to bigger (up to 150µm) and deeper vessels (around 900µm) that are visible in Fig. 13(d). Spurious horizontal lines are the result of motion artifacts.
Fig. 13 a: 3D rendering of microcirculation embedded in the structural data. 2x2mm cut at different depths to show the planar microcirculation network close to the epidermis/dermis transition and bigger vessels deeper in tissue. b-d: En-face projection at a depth of about 330µm over 50µm (b), 460µm over 200µm (c), and 880µm over 120µm (d). e: Fusion of b, c & d microcirculation in the R, G & B color channels respectively. Scale bar denotes 250µm in every picture.
Figure 13(e) is an attempt to visualize the 3D capillary network in a single picture using RGB fusion where the color channels encode the vessel structure of Fig. 13(b), 13(c) and 13(d) respectively.
4. Discussion
The xf-configuration shows convincing advantages compared to the standard illumination scheme. The axial beam confinement leads to a significantly enhanced imaging depth in scattering tissue such as skin. The self-reconstruction and dark field properties of the extended configuration are intrinsic features that have the potential to increase the quality of OCT pictures. The self-reconstruction property prevents image depth degradation from high absorbing or scattering sample features thanks to the ring illumination and seems also to play an important role for the reduction of multiple scattering and for the extension of the tissue axial imaging range. However, this assumption has to be further demonstrated by complementary experiments. The dark field effect on the other side did not have the same contrast enhancing effect for skin imaging as in microscopy. This is different in case of phase samples that exhibit only low scattering contrast [21]. It helps however to attenuate strong specular sample reflections, which is particularly useful when employing DAQ boards with limited dynamic range but high-speed sampling that are currently of strong interest for fast OCT applications. Indeed, it is not always possible to tilt the sample or to place it in contact with index matching components in order to avoid reflections.
The disadvantages of the extended focus reside in the presence of side lobes. First, energy is spread over different lobes, which results in a lower SNR at the confocal position compared to the standard case. However, the quality of xf-OCT en-face images and in particular the contrast of small structures (see Fig. 7) even in the vicinity of the focal region of the standard configuration seems much better than expected from the numerical SNR difference. Second, the side lobes affect the lateral resolution for depths distant from the detection confocal position. However the performance remains still superior to the standard case.
The presented imaging system is fully compatible with high-speed measurements offered by the FDML swept source technology. To demonstrate it, a buffer stage was connected to the light source output, allowing for an A-Scan rate multiplication [41]. A factor of 8 times, leading to 440 kA-Scan/s, was selected for comparison with the fundamental frequency of the swept source. Figure 14(a) and 14(b) compares single tomograms acquired at the A-Scan rate of 110 and 440kHz respectively. The previous 55kHz is equivalent to 110kHz in a SNR sense because the buffer stage produces copies of one sweep direction over which the power is spread. Of course, the acquisition time is decreased by a factor of 2. A small reduction in sensitivity, quantified to be around 4dB, is visible in the stratum corneum or at larger depths. The difference with theory can be explained by a slightly smaller spectral bandwidth as well as operation closer to shot-noise limit at 440kHz.
Fig. 14 Comparison of single tomograms acquired at the A-Scan rate of 110 and 440kHz with extended focus configuration in a and b respectively. Scale bar denotes 250µm.
The presented microcirculation contrasting method profits from the intrinsic advantages of the xf-configuration, in particular the reduction of the phase decorrelation effect leading to reduced vessel tails in the variance image. It is possible to improve the axial definition of vessels further in post-processing, e.g. by applying a step down exponential filter before a 3D vessel tracing algorithm based on cylindroidal superellipsoids fit [24].
The rapid development and application of high-speed swept sources open exciting perspectives for the assessment of capillary networks in healthy and diseased tissue. It is now possible to maintain a proper lateral sampling while increasing the fast scanning frequency giving access to phase difference information between tomograms. This provides not only qualitative information as it is shown in the present work, but also quantitative flow measurements. Phase resolved data should ease the vessels segmentation because of the deterministic relation between vessel pixels compared to random phase values that result from multiple scattering. Finally, high-speed makes 4D imaging possible, which will open new perspectives in the characterization and understanding of dynamic processes.
5. Conclusion
An extended focus OCT (xf-OCT) imaging system using a Bessel illumination was implemented in order to improve the lateral resolution while preserving the depth imaging range and the advantages of FDOCT. Compared to a standard system and applied to skin in-vivo imaging, it exhibits a larger axial imaging range or a better lateral resolution depending on the choice of the detection NA. Moreover, it is shown how OCT imaging can be improved thanks to the self-reconstruction property of Bessel beams. The decoupling between illumination and detection enables dark field imaging, which is interesting to attenuate the sample first specular reflection and permits an efficient use of limited dynamic range 8 bit high-speed acquisition card. The system is applied to functional imaging for the visualization of micro-vasculature in-vivo of the skin. The Bessel illumination, compared to the standard case, suffers less from decorrelation artifacts below blood vessels and thus offers a better vessel segmentation. Finally, the capability of the system to perform high-speed imaging offered by the FDML swept source technology is demonstrated.
Acknowledgment
We acknowledge financial support from the European Commission Seventh Framework Programme (FP7) HEALTH program (grant 201880, FUN OCT), as well as kind support for the FDML source from Thomas Klein.
References and links
1. W. Drexler and J. G. Fujimoto, eds., Optical Coherence Tomography - Technology and Applications (Springer, 2008).
2. J. M. Schmitt, S. L. Lee, and K. M. Yung, “An optical coherence microscope with enhanced resolving power in thick tissue,” Opt. Commun. 142(4-6), 203–207 (1997). [CrossRef]
3. A. G. Podoleanu, G. M. Dobre, and D. A. Jackson, “En-face coherence imaging using galvanometer scanner modulation,” Opt. Lett. 23(3), 147–149 (1998). [CrossRef]
4. M. Pircher, B. Baumann, E. Götzinger, H. Sattmann, and C. K. Hitzenberger, “Simultaneous SLO/OCT imaging of the human retina with axial eye motion correction,” Opt. Express 15(25), 16922–16932 (2007). [CrossRef] [PubMed]
5. A. F. Fercher, C. K. Hitzenberger, G. Kamp, and S. Y. El-Zaiat, “Measurement of intraocular distances by backscattering spectral interferometry,” Opt. Commun. 117(1-2), 43–48 (1995). [CrossRef]
6. M. Wojtkowski, R. Leitgeb, A. Kowalczyk, T. Bajraszewski, and A. F. Fercher, “In vivo human retinal imaging by Fourier domain optical coherence tomography,” J. Biomed. Opt. 7(3), 457–463 (2002). [CrossRef] [PubMed]
7. R. Leitgeb, C. K. Hitzenberger, and A. F. Fercher, “Performance of fourier domain vs. time domain optical coherence tomography,” Opt. Express 11(8), 889–894 (2003). [CrossRef] [PubMed]
8. M. Choma, M. Sarunic, C. Yang, and J. Izatt, “Sensitivity advantage of swept source and Fourier domain optical coherence tomography,” Opt. Express 11(18), 2183–2189 (2003). [CrossRef] [PubMed]
9. J. F. de Boer, B. Cense, B. H. Park, M. C. Pierce, G. J. Tearney, and B. E. Bouma, “Improved signal-to-noise ratio in spectral-domain compared with time-domain optical coherence tomography,” Opt. Lett. 28(21), 2067–2069 (2003). [CrossRef] [PubMed]
10. R. Huber, M. Wojtkowski, J. G. Fujimoto, J. Y. Jiang, and A. E. Cable, “Three-dimensional and C-mode OCT imaging with a compact, frequency swept laser source at 1300 nm,” Opt. Express 13(26), 10523–10538 (2005). [CrossRef] [PubMed]
11. T. S. Ralston, D. L. Marks, F. Kamalabadi, and S. A. Boppart, “Deconvolution methods for mitigation of transverse blurring in optical coherence tomography,” IEEE Trans. Image Process. 14(9), 1254–1264 (2005). [CrossRef] [PubMed]
12. Y. Yasuno, Y. Sando, J. I. Sugisaka, T. Endo, S. Makita, G. Aoki, M. Itoh, and T. Yatagai, “In-focus Fourier-domain optical coherence tomography by complex numerical method,” Opt. Quantum Electron. 37(13-15), 1185–1189 (2005). [CrossRef]
13. Y. Yasuno, J. I. Sugisaka, Y. Sando, Y. Nakamura, S. Makita, M. Itoh, and T. Yatagai, “Non-iterative numerical method for laterally superresolving Fourier domain optical coherence tomography,” Opt. Express 14(3), 1006–1020 (2006). [CrossRef] [PubMed]
14. L. F. Yu, B. Rao, J. Zhang, J. P. Su, Q. Wang, S. G. Guo, and Z. P. Chen, “Improved lateral resolution in optical coherence tomography by digital focusing using two-dimensional numerical diffraction method,” Opt. Express 15(12), 7634–7641 (2007). [CrossRef] [PubMed]
15. V. X. Yang, N. Munce, J. Pekar, M. L. Gordon, S. Lo, N. E. Marcon, B. C. Wilson, and I. A. Vitkin, “Micromachined array tip for multifocus fiber-based optical coherence tomography,” Opt. Lett. 29(15), 1754–1756 (2004). [CrossRef] [PubMed]
16. J. Holmes, S. Hattersley, N. Stone, F. Bazant-Hegemark, and H. Barr, “Multi-channel Fourier domain OCT system with superior lateral resolution for biomedical applications,” Proc. SPIE 6847, 684700 (2008).
17. Z. H. Ding, H. W. Ren, Y. H. Zhao, J. S. Nelson, and Z. P. Chen, “High-resolution optical coherence tomography over a large depth range with an axicon lens,” Opt. Lett. 27(4), 243–245 (2002). [CrossRef]
18. K. S. Lee and J. P. Rolland, “Bessel beam spectral-domain high-resolution optical coherence tomography with micro-optic axicon providing extended focusing range,” Opt. Lett. 33(15), 1696–1698 (2008). [CrossRef] [PubMed]
19. R. M. Herman and T. A. Wiggins, “Production and uses of diffractionless beams,” J. Opt. Soc. Am. A 8(6), 932 (1991). [CrossRef]
20. R. A. Leitgeb, M. Villiger, A. H. Bachmann, L. Steinmann, and T. Lasser, “Extended focus depth for Fourier domain optical coherence microscopy,” Opt. Lett. 31(16), 2450–2452 (2006). [CrossRef] [PubMed]
21. M. Villiger, C. Pache, and T. Lasser, “Dark-field optical coherence microscopy,” Opt. Lett. 35(20), 3489–3491 (2010). [CrossRef] [PubMed]
22. T. Schmoll, C. Kolbitsch, and R. A. Leitgeb, “Ultra-high-speed volumetric tomography of human retinal blood flow,” Opt. Express 17(5), 4166–4176 (2009). [CrossRef] [PubMed]
23. I. Grulkowski, I. Gorczynska, M. Szkulmowski, D. Szlag, A. Szkulmowska, R. A. Leitgeb, A. Kowalczyk, and M. Wojtkowski, “Scanning protocols dedicated to smart velocity ranging in spectral OCT,” Opt. Express 17(26), 23736–23754 (2009). [CrossRef]
24. B. J. Vakoc, R. M. Lanning, J. A. Tyrrell, T. P. Padera, L. A. Bartlett, T. Stylianopoulos, L. L. Munn, G. J. Tearney, D. Fukumura, R. K. Jain, and B. E. Bouma, “Three-dimensional microscopy of the tumor microenvironment in vivo using optical frequency domain imaging,” Nat. Med. 15(10), 1219–1223 (2009). [CrossRef] [PubMed]
25. A. Mariampillai, B. A. Standish, E. H. Moriyama, M. Khurana, N. R. Munce, M. K. Leung, J. Jiang, A. Cable, B. C. Wilson, I. A. Vitkin, and V. X. Yang, “Speckle variance detection of microvasculature using swept-source optical coherence tomography,” Opt. Lett. 33(13), 1530–1532 (2008). [CrossRef] [PubMed]
26. R. Leitgeb, L. Schmetterer, M. Wojtkowski, C. Hitzenberger, M. Sticker, and A. Fercher, “Flow Velocity Measurements by Frequency Domain Short Coherence Interferometry,” Proc. SPIE 4619, 16–21 (2002). [CrossRef]
27. R. A. Leitgeb, L. Schmetterer, C. K. Hitzenberger, A. F. Fercher, F. Berisha, M. Wojtkowski, and T. Bajraszewski, “Real-time measurement of in vitro flow by Fourier-domain color Doppler optical coherence tomography,” Opt. Lett. 29(2), 171–173 (2004). [CrossRef] [PubMed]
28. N. Nassif, B. Cense, B. H. Park, S. H. Yun, T. C. Chen, B. E. Bouma, G. J. Tearney, and J. F. de Boer, “In vivo human retinal imaging by ultrahigh-speed spectral domain optical coherence tomography,” Opt. Lett. 29(5), 480–482 (2004). [CrossRef] [PubMed]
29. B. Vakoc, S. Yun, J. de Boer, G. Tearney, and B. Bouma, “Phase-resolved optical frequency domain imaging,” Opt. Express 13(14), 5483–5493 (2005). [CrossRef] [PubMed]
30. S. Makita, Y. Hong, M. Yamanari, T. Yatagai, and Y. Yasuno, “Optical coherence angiography,” Opt. Express 14(17), 7821–7840 (2006). [CrossRef] [PubMed]
31. J. Fingler, D. Schwartz, C. Yang, and S. E. Fraser, “Mobility and transverse flow visualization using phase variance contrast with spectral domain optical coherence tomography,” Opt. Express 15(20), 12636–12653 (2007). [CrossRef] [PubMed]
32. Y. K. Tao, A. M. Davis, and J. A. Izatt, “Single-pass volumetric bidirectional blood flow imaging spectral domain optical coherence tomography using a modified Hilbert transform,” Opt. Express 16(16), 12350–12361 (2008). [CrossRef] [PubMed]
33. M. Szkulmowski, A. Szkulmowska, T. Bajraszewski, A. Kowalczyk, and M. Wojtkowski, “Flow velocity estimation using joint Spectral and Time domain Optical Coherence Tomography,” Opt. Express 16(9), 6008–6025 (2008). [CrossRef] [PubMed]
34. L. An, J. Qin, and R. K. Wang, “Ultrahigh sensitive optical microangiography for in vivo imaging of microcirculations within human skin tissue beds,” Opt. Express 18(8), 8220–8228 (2010). [CrossRef] [PubMed]
35. A. Alex, B. Považay, B. Hofer, S. Popov, C. Glittenberg, S. Binder, and W. Drexler, “Multispectral in vivo three-dimensional optical coherence tomography of human skin,” J. Biomed. Opt. 15(2), 026025 (2010). [CrossRef] [PubMed]
36. R. Huber, M. Wojtkowski, and J. G. Fujimoto, “Fourier Domain Mode Locking (FDML): A new laser operating regime and applications for optical coherence tomography,” Opt. Express 14(8), 3225–3237 (2006). [CrossRef] [PubMed]
37. S. W. Huang, A. D. Aguirre, R. A. Huber, D. C. Adler, and J. G. Fujimoto, “Swept source optical coherence microscopy using a Fourier domain mode-locked laser,” Opt. Express 15(10), 6210–6217 (2007). [CrossRef] [PubMed]
38. W. Wieser, B. R. Biedermann, T. Klein, C. M. Eigenwillig, and R. Huber, “Multi-megahertz OCT: High quality 3D imaging at 20 million A-scans and 4.5 GVoxels per second,” Opt. Express 18(14), 14685–14704 (2010). [CrossRef] [PubMed]
39. F. O. Fahrbach, P. Simon, and A. Rohrbach, “Microscopy with self-reconstructing beams,” Nat. Photonics 4(11), 780–785 (2010). [CrossRef]
40. B. D. Goldberg, B. J. Vakoc, W. Y. Oh, M. J. Suter, S. Waxman, M. I. Freilich, B. E. Bouma, and G. J. Tearney, “Performance of reduced bit-depth acquisition for optical frequency domain imaging,” Opt. Express 17(19), 16957–16968 (2009). [CrossRef] [PubMed]
41. R. Huber, D. C. Adler, and J. G. Fujimoto, “Buffered Fourier domain mode locking: Unidirectional swept laser sources for optical coherence tomography imaging at 370,000 lines/s,” Opt. Lett. 31(20), 2975–2977 (2006). [CrossRef] [PubMed]
