User:Ariceggers/sandbox

= Dual Axis Optical Coherence Tomography =

Dual-axis optical coherence tomography (DA-OCT) which enables deep tissue imaging by using a novel off-axis illumination/detection configuration. DA-OCT offers a 100-fold speed increase compared to its predecessor, multispectral multiple-scattering low coherence interferometry (ms2/LCI), by using a new beam scanning mechanism based on a microelectrical- mechanical system (MEMS) mirror. The data acquisition scheme was altered to take advantage of this scanning speed, producing tomographic images at a rate of 4 frames (B-scans) per second. DA-OCT differs from ms2/LCI in that the dual axes intersect at a shallower depth (~1mm). This difference, coupled with the faster scanning speed shifts the detection priority from multiply scattered to ballistic light. The utility of this approach was demonstrated by imaging both ex vivo porcine ear skin and in vivo rat skin from a Macfarlane flap model. The enhanced penetration depth provided by the DA-OCT system will be beneficial to various clinical applications in dermatology and surgery.

Optical Design
The DA-OCT system, shown in Fig. 2, is similar to our previous ms2/LCI setup [2]. Light from a supercontinuum laser source (Fianium SC-450-4) is filtered with a near infrared (NIR) bandpass filter (795/150 nm Brightline bandpass filter, Semrock, Rochester, NY) before entering a Mach–Zehnder interferometer. Two 90:10 beamsplitter cubes are used to generate interference between the sample and reference arms. A magnified view of the imaging end of the DA-OCT system is shown in the insets of Fig. 2. A 100 mm focal length lens (L1) is used to focus light onto the sample, producing a spot size measured to be 27.3 μm. However, the lateral resolution achieved in practice will be degraded due to the presence of scattering. The incident power on the sample can be adjusted with a variable neutral density (ND) filter. The input beam to the sample is offset from the optical axis of the lens by 3°.

While larger crossing angles may improve the rejection of out-of-focus background [6], tighter spatial rejection will lead to a smaller imaging range, which is undesirable for deep imaging. Besides, larger offset angles also lead to greater signal attenuation due to longer photon paths within the tissue. Therefore, this offset angle of 3° is selected to optimize SNR for deep imaging [3 ]. Lateral beam scanning is performed using the aforementioned MEMS mirror (Mirrorcle Technologies, Richmond, CA).

Light returned by the sample is collected and interfered with the reference arm before being delivered to a customized spectrometer. The spectrometer is centered at 770 nm with 108 nm bandwidth. A 1400 lines/mm grating (T-1400-800s, LightSmyth, Eugene, OR) spectrally disperses the collected light onto a 12-bit CCD camera (Piranha HS-40, Teledyne Dalsa, Waterloo, Ontario, Canada). The camera has a sensor of 4096 × 96 pixels with a pixel size of 7 × 7 μm. The maximum line rate of the camera is 18 kHz. The pixel-limited spectral resolution of the spectrometer is 26.4 pm, which determined the theoretical depth range of the DA-OCT system to be 5.7 mm [7]. However, the true spectral resolution is approximately 74 pm based on the measured 6 dB roll-off of z6 dB 1.8 mm. The SNR of the system was measured by placing a mirror on the sample plane and calculating the ratio of the amplitude of the peak signal to the standard deviation of the noise. The maximum SNR of a single A-scan was measured to be 101 dB.

Axial Resolution
The use of an off-axis design in DA-OCT means that the axial resolution of the system is not simply defined by calculating the coherence length of the source. Instead, both the optical properties of the tissue and the beam geometry influence the axial resolution. Scattering in the sample produces a distribution of photon path lengths, thus blurring axial features and leading to degraded resolution. The amount of resolution degradation due to scattering depends on the number of scattering events and the optical properties of the sample, including scattering MFP and anisotropy. The use of the off-axis configuration also influences the collection of multiply scattered photons and modifies the relationship between depth of focus (DOF) and numerical aperture (NA) that governs the imaging performance of an on-axis configuration. The beam geometry particularly influences depth resolution in the presence of scattering when the depth resolution has been degraded the significantly beyond that afforded by the coherence gate. To be clear, path length resolution is preserved by the coherence gate, but it may span multiple depth layers due to scattering. A more detailed discussion of this relationship can be found in our previous study [8]. Therefore, although the coherence length is calculated to be approximately 3 μm, the effective axial resolution of this DA-OCT geometry can be as large as 100 μm at its maximum penetration depth [3].

Acquisition Speed
A major improvement of the DA-OCT system over its ms2/LCI predecessor is the increase in acquisition speed. The imaging speed of the original system was limited by the sample translation speed and the need to average multiple A-scans to detect multiply scattered light. However, the low acquisition speed was problematic for in vivo measurements where movement caused blurring artifacts [3]. Therefore, previous studies with the ms2/LCI system were limited to ex vivo samples, and a relatively slow camera acquisition time of ∼5.3 ms was used for a single point measurement (A-scan) with multiple A-scans averaged to trade for higher SNR [3]. The DA-OCT system avoids this problem by introducing a beam-scanning mechanism using a MEMS mirror. This allows us to operate the area camera in a line scan configuration, reading out one single line of 4 k pixels from the CCD instead of a full area readout so that the full acquisition speed of the camera is used to maximize the A-scan rate. This would not have improved the performance of the previous system because it required sample translation. The acquisition speed for a single point scan has been reduced to approximately 56 μs, almost 100 times faster than the previous ms2/LCI system.

Thin Tissue Classification
The penetration depth of DA-OCT is shown by imaging ex vivo porcine ear skin tissue. It was previously demonstrated that pig ear skin, as compared with other regions of the pig skin, exhibits the best resemblance to human skin and therefore is the optimal ex vivo model for human skin studies [9]. Figure 3 shows the representative images of porcine ear skin acquired from both our DA-OCT and a commercial OCT system with a similar center wavelength and bandwidth (850/155 nm, Spark DRC, Wasatch Photonics Inc., Durham, NC).

A notable aspect of the DA-OCT image is the brightness of the epidermis compared with the other features. At this depth, the illumination and detection pathways are separated by ∼75 μm, and thus no direct photon pathways are possible for single backscattering or reflected light. Instead, we theorize that photons can travel laterally in this layer and produce a strong signal. Indeed, the thickness of this layer appears larger than in the conventional OCT image, suggesting that multiply scattered photons are detected. This effect is not apparent in deeper tissue structures. Overall, an improved penetration depth of ∼1 mm can be realized by the off-axis configuration compared with the 400–500 μm for conventional OCT measurements [10].

Tissue Viability
In this study, we evaluate the ability of DA-OCT to assess tissue viability using a rat McFarlane flap model. Skin flaps are often used to model complex surgical defects. Flap viability can be evaluated by monitoring tissue perfusion post-operatively where early identification of compromised perfusion can improve flap salvage [11-13]. Therefore, a noninvasive technology for measuring the spatial distribution of tissue optical properties as a biomarker of flap viability can have significant utility.

Using the method described in the previous section, brel is calculated pixel-wise for each of the 6 positions with 125 μm depth intervals for day 1, 4 and 7. The data were then processed by a spline interpolation algorithm before being color-coded based on the interpolated brel values. The color-coded brel maps are shown in Fig.5. It can be observed that brel generally decreases in all positions and depths from day 1 to day 7, which potentially suggests that the severity of tissue injury increases with time postoperatively. More specifically, the decrease in brel started at superficial depths of position 5 and 6 on day 1. However, tissue towards the end of the flap and at deeper depths still yields a relatively high brel, which indicates healthy tissue. On day 4, brel started to decrease both along the length of the flap and vertically towards deeper depths. The severity of tissue injury increased dramatically on day 7, by that time tissue at position 3-6 yields brel well below 0.5 for all depths. The fact that brel is relatively low even at the deepest depths that the current ms2/LCI system can penetrate potentially



In conclusion, depth-resolved spectroscopic deep tissue imaging was used to provide non-invasive flap viability assessment in an animal model. Changes in the relative power-law exponent brel were found to correlate with the severity of the injury, which also matches the gradient in tissue perfusion along the length of the flap. These results suggest that the approach can be used for the evaluation of skin injuries at extended depths compared to conventional optical modalities. Extension to a rapid scanning system has opened the door for in vivo applications.