An improved image processing procedure for suppressing the phase noise due to a motion artifact acquired during optical coherence tomography scanning and effectively illustrating the blood vessel distribution in a living tissue is demonstrated. This new processing procedure and the widely used procedure for micro-angiography application are based on the selection of high-frequency components in the spatial-frequency spectrum of B-mode scanning (x-space), which are contributed from the image portions of moving objects. However, by switching the processing order between the x-space and k-space, the new processing procedure shows the superior function of effectively suppressing the phase noise due to a motion artifact. After the blood vessel positions are precisely acquired based on the new processing procedure, the projected blood flow speed can be more accurately calibrated based on a previously reported method. The demonstrated new procedure is useful for clinical micro-angiography application, in which a stepping motor of generating motion artifacts is usually used in the scanning probe.
© 2011 OSA
Because it can provide important information of blood flow in living bio-tissue, optical Doppler tomography (ODT) has become an attractive technique in the development of the optical coherence tomography (OCT) technology [1–11]. In processing ODT scanning images, the suppression of the phase noise caused by motion artifacts is an important issue. The phase noise may seriously degrade the processed ODT images. A few methods have been proposed for suppressing such phase noise. For instance, because the phase noise in an A-mode scan caused by a motion artifact is usually quite uniform, the phase shift value of maximum counts among the pixels in an A-mode scan can be found and is subtracted from the original values to suppress the effect of motion artifact . The search for the phase shift value of maximum counts can be affected by the blood flow in the sample. Such an effect is small if the blood vessel covers only a small range in an A-mode scan. However, when a large-cross-section blood vessel appears in an OCT image, this approach for suppressing phase noise may become ineffective. In an alternative approach, phase noise can be suppressed by B-mode scanning the sample twice of reversed directions . The phase variations due to blood flow in these two scans are the same; however, the background phase noise values in these two scans are mutually reversed in sign. Therefore, the background phase noise can be suppressed at the expense of increasing scanning time.
On the other hand, without the knowledge of blood flow speed, the information of the blood vessel distribution in a living tissue is useful for understanding the tissue condition, particularly around a tumor [9–11]. For such micro-angiography applications, OCT scanning and the related image processing techniques have been shown to be quite useful. The optical micro-angiography (OMAG) represents a powerful technique for image processing [8, 14–18]. In this technique, the data in the k-x image is Fourier transformed along the x direction (the B-mode scan direction) to obtain spatial-frequency spectrum for each k value. In such a spectrum, the high- and low-frequency components correspond to the contributions of blood flow and background structure, respectively, in the tissue sample. By deleting a reasonable range of low-frequency components, the blood flow information can be extracted and the low-frequency phase noise can be suppressed. The OMAG technique has been proved to be quite useful in many applications of micro-angiography [14–17]. However, the use of this technique becomes limited when high-frequency phase noise exists to mix with the contribution of blood flow. Such phase noise may originate from the light source (such as a swept source), sample motion, system instability, and system scanning mechanisms (such as the scanning nature of a stepping motor). The phase noise must be effectively suppressed for obtaining high-quality micro-angiography images.
In this paper, we demonstrate an alternative imaging processing procedure for effectively suppressing the high-frequency phase noise due to a motion artifact. The basic concept of the proposed image processing procedure is similar to the OMAG technique. However, the image processing procedure is modified. With this new procedure, we can effectively suppress the phase noise due to the scanning nature of a stepping motor, which is widely used for building an OCT scanning probe. Such a probe has been used for scanning oral cavity to early diagnose oral cancer and precancer [19, 20]. Although the method of subtracting the maximum-count phase shift in each A-mode scan has been reported for suppressing the phase noise due to motion artifacts , in this paper, we illustrate the problem of using this method when the blood vessel distribution covers a large cross section in an OCT image. We will show that by using our processing procedure for obtaining the micro-angiography information in advance, we can better use the maximum-count phase-shift method for acquiring more accurate ODT results. In section 2 of this paper, the used OCT systems for in vivo scanning and the basic scanning results are shown. The theory behind our approach is discussed in section 3. Then, the processed images and their comparisons with those based on the conventional techniques are presented in section 4. Finally, conclusions are drawn in section 5.
2. Optical coherence tomography systems and scanning results
Figure 1(a) shows the common setup of the two used swept-source OCT (SS-OCT) systems in this study. In the setup, besides the fiber Mach-Zehnder interferometer to form the major body of the OCT system, 5% of the light source power is sent to a fiber Bragg grating through a circulator for synchronizing the OCT signal acquisition with the frequency sweeping of the swept source . The interfered spectral signals are monitored by a balanced detector (Thorlabs, PDB12A) and acquired by a personal computer (PC) through a data acquisition (DAQ) card (National Instruments, PCI-5122 or PXIe-5122). All the fiber couplers and circulators in the systems are manufactured by the company of Thorlabs. In the sample arm, one of the OCT systems is connected to a scanning probe containing a stepping motor. The other is connected to a scanning galvanometer. In the OCT system with the scanning probe, the swept source has a sweeping wavelength range of 110 nm (Santec, HSL2000). In the other OCT system with the scanning galvanometer, the swept source has a sweeping wavelength range of 170 nm (Santec, HSL2100). The central wavelength and scanning rate of both light sources are 1310 nm and 20 kHz, respectively. The fiber Bragg grating (Lead Fiber Optics Co.) used in the OCT system of probe scanning has the Bragg wavelength at 1260 nm. That used (also Lead Fiber Optics Co.) in the OCT system of galvanometer scanning has the Bragg wavelength at 1230 nm. Figure 1(b) shows the schematic drawing of the scanning probe. Here, the linear stepping motor (Montrol System Co., A35H4N-24) controls the translational motion of the probe shaft for B-mode scan to achieve a scanning speed of 10 cm/s and a 1 cm scanning length. However, in this study, we use the scanning speed of only 1.25 cm/s with a B-mode scanning range of 1.25 mm. The probe has a length of 10 cm beyond the stepping motor at the proximal end. The square geometry of the probe cross section has the dimension of 0.8 cm. At the distal end of the probe, an opening of 2.5 cm x 0.5 cm is fabricated for light illumination onto a sample and backscattered light collection. The illuminating light beam is focused by a Grin lens (OZ Optics) at the fiber end. Its propagation direction is changed by the reflection of a prism mirror (OZ Optics) inside the probe. Figure 1(c) shows the setup of the galvanometer scanning unit. In this setup, the light beam is swung for B-mode scan through the reflection by a mirror mounted on a galvanometer (Thorlabs, GVS002). The swung light beam is then focused by an objective lens (Mitutoyo, Q73014308A) for illuminating a sample. The OCT system with the scanning probe has been used for clinical scanning of oral cancer patients in a hospital and has led to the discovery of several effective indicators for the diagnoses of oral cancer and precancer [19, 20]. It is noted that in fabricating a scanning probe for clinical application, optical micro-electro-mechanical systems (MEMSs) have been used for two-dimensional or three-dimensional lateral scans [22, 23]. However, the MEMS technology does not seem mature enough for the application of fabricating OCT scanning probes. In practice, a stepping motor is still one of the most reliable components for B-mode scanning of a probe. However, because of the stepping motion nature of such a component, quasi-periodical phase noise is added to OCT signal during the B-mode scan.
In Fig. 2(a) , we show an OCT image of human buccal mucosa of a healthy volunteer through probe scanning. To acquire a more reliable time-resolved phase variation at the same lateral location based on the difference between two neighboring A-mode scans, the B-mode scan pixel size is designed to be quite small. The A- and B-mode scan pixel sizes are set at 7.8 and 0.63 μm, respectively, in the OCT system with probe scanning. Also, the A- and B-mode scan pixel sizes are set at 3.18 and 0.4 μm, respectively, in the OCT system with galvanometer scanning. The image in Fig. 2(a) consists of 190 and 2000 pixels in the A- and B-mode scan directions, respectively. Here, the epithelium (EP) and lamina propria (LP) layers can be clearly observed. Also, the plastic plate (PP) for protecting the optical components inside the probe from contamination can be seen. The mapping of the phase shift between two neighboring A-mode scans of this OCT image is shown in Fig. 2(b). Here, one can see the quasi-periodical phase variation along the B-mode scan direction. Such phase noise may mix with the signal produced by blood flow in the tissue such that the calibration of blood vessel distribution becomes difficult. To further illustrate that the phase shift distribution shown in Fig. 2(b) is caused by OCT system operation, instead of sample structure, we use the same OCT system (with the probe) to scan a sample consisting of 14 layers of Scotch tape on a coverslip. Figures 3(a) and 3(b) show the OCT structure image and phase shift distribution, respectively. Also, the variation of the maximum-count phase shift in each A-mode scan along the B-mode scan (the x direction) is shown in Fig. 3(c). Here, one can see the similar quasi-periodic variation of phase shift, clearly indicating the phase noise caused by the system-related (stepping motor) motion artifacts.
The basic idea for mapping the blood vessel distribution relies on the faster time-resolved variation of OCT signal at those pixels with object motions. When the B-mode scan pixel size is small enough, the lateral variation of OCT intensity signal at a location of object motion will show a feature of higher spatial frequency, when compared with other locations of static structures. In this situation, the high-frequency components of a spatial-frequency spectrum, obtained after a Fourier transform of the OCT intensity signal along the B-mode scan direction, correspond to the contributions from those locations of object motions. A reasonable selection of the high-frequency components of the spatial-frequency spectrum can provide us with the blood vessel mapping after an inverse Fourier transform. This basic concept has been used in the OMAG technique. However, this technique becomes less effective when the condition of motion artifact or phase noise like the case shown in Fig. 2(b) is encountered. To demonstrate such a situation, we use the OCT system with galvanometer scanning to scan human skin on one of the fingers of a volunteer. During the scanning, the volunteer intentionally moved the finger to produce a motion artifact. Figures 4(a) and 4(b) show the OCT structure image and its phase shift mapping, respectively. The image in Fig. 4(a) consists of 285 and 1250 pixels in the A- and B-mode scan directions, respectively. Although the effect of the motion artifact cannot be identified in the structure image, it can be clearly observed in the phase shift mapping. In Fig. 4(b), a signal feature, indicated by the arrow, corresponding to a blood vessel, can be clearly observed. The phase noise caused by finger motion can be seen in the right portion. Such phase noise masks another blood vessel feature, as to be seen in the following discussion.
The different lateral signal variations between the image locations of moving objects and static structures can be seen in Figs. 5(a) -5(d). Here, Fig. 5(a) duplicates Fig. 4(a) with a rectangular region (circled by the green dashed lines) being magnified to give Fig. 5(b). Near the center of Fig. 5(b), one can see a region of fast signal variation along the lateral direction. By plotting a horizontal blue dashed line to pass this region and a red dashed line above it (outside this region) in Fig. 5(b), one can clearly see the different lateral variations in their line-scan profiles, as demonstrated in Figs. 5(c) and 5(d) for the upper (red) and lower (blue) line scans, respectively. Here, one can see that in the lateral range x between 40 and 110 μm of Fig. 5(d), the signal variation is faster than those outside this x range and all the x range in Fig. 5(c). The fast signal variation is caused by the blood cell motion in the blood vessel leading to structure changes with time. Therefore, in successive A-mode scans, fast-varying backscattering signals are recorded. The portion of fast variation in Fig. 5(d) contributes to the high-frequency components in the lateral spatial-frequency spectrum. The selection of those high-frequency components for inverse Fourier transform can provide us with blood vessel image.
Mathematically, if we start with the real interfered spectral signals collected by an OCT system at a particular depth zl, A(k, zl, x), and assume that there is no motion artifact or phase noise, the image processing procedure in the OMAG technique for obtaining the blood vessel distribution signal, B(zl, x), is as follows:
Here, and represent the Fourier transform and inverse Fourier transform, respectively, along the x direction. The notation, , stands for the Fourier transform in the k domain (frequency). Also, the notation W denotes a window function for deleting the low-frequency components in both real and imaginary parts of the spatial-frequency spectrum. Meanwhile, S represents the mirror image suppression operation. In addition, Abs stands for the mathematical operation of taking the absolute value. It is noted that z and x have been used for denoting the coordinates in the A- and B-mode scanning directions, respectively. The image processing procedure described in Eq. (1) (referred to as the reference procedure) has been widely used for micro-angiography applications. However, it is difficult to obtain clear blood vessel distributions based on this procedure when motion artifacts exist. A motion artifact can usually produce phase noise, ϕ(zl, x), in A(k, zl, x) with laterally high-spatial-frequency components. In this situation, the OCT signal can be expressed as
Here, a(zl, x) stands for the backscattering signal from the location of (zl, x). The phase noise cannot be removed in the whole mathematical operation procedure in Eq. (1) such that it may blur the blood vessel image. Nevertheless, a change of the mathematical operation procedure in Eq. (1) can help in deleting the phase noise.
Our proposed image processing procedure is as follows:
Here, after the mirror image suppression operation (S), the absolute-value signal becomes
Therefore, the phase noise is removed. However, in the reference procedure shown in Eq. (1), after the mirror image suppression operation (S), the absolute-value signal becomes
Here, the notation stands for a convolution operation. With the convolution operation and window function, the phase noise cannot be suppressed even after we take the absolute value. Therefore, with the reference processing procedure, the phase noise will blur the blood vessel image. It is noted that if ϕ(zl, x) represents a constant phase shift, which is commonly used in ODT, it can be deleted through the reference procedure. Also, if the motion artifact produces only the phase noise of low spatial frequency components, it can be suppressed by the window function. In this situation, the reference processing procedure should also work well for suppressing the phase noise.
The phase shift mapping images like Figs. 2(b) and 4(b) can be used for evaluating the blood speed component projected on the image plane to obtain the ODT images. When such an image contains phase noise due to a motion artifact, a method has been proposed to suppress the phase noise before the evaluation of projected blood speed component . In this method, an A-mode scan range of a two-dimensional OCT image is selected for searching the phase shift value of maximum counts in each A-mode scan. Then, this maximum-count phase shift value is subtracted from all the phase shift values in this A-mode scan. Because the phase noise due to a motion artifact is roughly a constant along an A-mode scan, such a subtraction operation can suppress the noise. However, when an A-mode scan covers a blood vessel of a large cross section, the phase shift due to blood flow may disturb the accurate search for the maximum-count phase shift value such that errors may exist in evaluating the blood flow speed. In our approach, based on the aforementioned procedure, we can first accurately obtain the locations (depths) of the blood vessels in an image. Therefore, in selecting the depth range for searching the phase-shift value of maximum counts, we can avoid the depths of blood vessels for reducing the errors caused by the phase shifts of blood flow. Hence, our procedure for the accurate determination of blood vessel positions first is useful for more accurately evaluating the projected blood flow speeds.
4. Scanning and processing results
Figures 6(a) and 6(b) show the processed results, i.e., B(zl, x), based on our procedure and the reference procedure, respectively, of the scanning result shown in Fig. 4(a). Here, one can see that with our processing procedure, two blood vessels can be clearly seen, as indicated by the arrows. However, with the reference processing procedure, the blood vessel on the right is masked by the vertical stripes caused by the intentionally produced motion artifact. Similar comparison can be seen in Figs. 7(a) and 7(b), which show the processed results based on our procedure and the reference procedure, respectively, of the mucosa image in Fig. 2(a). Again, our processing procedure is superior to the reference one in acquiring the blood vessel distribution. At least six blood vessels can be identified in Fig. 7(a), as indicated by arrows and circles. It is noted that if the phase noise distributions in Fig. 2(b) and 4(b) can be effectively suppressed, say, through the method of subtracting the maximum-count phase shift in each A-mode scan, the reference procedure can also be used for obtaining clear blood vessel mapping. However, as far as the micro-angiography application is concerned, the required extra work of suppressing the phase noise due to motion artifacts before applying the reference image processing procedure represents a drawback. Actually, after suppressing the phase noise, we can obtain the ODT image. The ODT image contains the information of blood vessel distribution (micro-angiography) and projected blood flow speed. The suppressions of phase noise and the ODT image illustrations of the OCT results in Figs. 2 and 4 will be discussed later with Figs. 8 -11 .
To demonstrate the phase noise suppression and the evaluation of the projected blood flow speeds in the blood vessels, we duplicate Fig. 6(a) as Fig. 8(a) with three selections of A-mode scan depth range as I, II, and III. Each selected depth range is used to search the phase-shift value of maximum counts in every A-mode scan for suppressing the phase noise caused by the motion artifact. The obtained ODT images based on selected ranges I, II, and III are shown in Figs. 8(b), 8(c), and 8(d), respectively. Here, one can see that with selected range I, which does not cover any blood vessel, the ODT image is quite clear. On the other hand, with selected range II, which covers the two blood vessels in this image, noise still exists and the reading of blood flow speed can be disturbed. Nevertheless, with selected range III, which covers the whole depth range of significant OCT signal strength, the ODT image looks as clear as that with selected range I. Because of the selected large depth range, the effect of the phase shift caused by the blood flow of a small cross section has a less significant contribution to the result in searching the phase-shift of maximum counts. Therefore, as long as the selected depth range is large compared to the blood vessel cross section, the prior knowledge of blood vessel location becomes less important. The projected blood flow speed distributions in the blood vessels shown in Figs. 8(b)-8(d) are unclear due to the limitation of image resolution. To clearly demonstrate the flow speed distributions, we plot the line-scan profiles along the horizontal dashed lines in Figs. 8(b)-8(d) to give the spiky (red) curves (labeled by “Data”) in Figs. 9(a) -9(c), respectively. Here, only the 250-μm ranges on the right in Figs. 8(b)-8(d) are shown to illustrate the two blood vessels (indicated by the two arrows). Also, to clearly show the variation of projected blood flow speed along the B-mode scan direction, a smoother curve (labeled by “Fitting”) is plotted after the high spatial-frequency components are filtered in each of Figs. 9(a)-9(c). We can see that in either case of Figs. 9(a) and 9(c), in which phase noise is better suppressed, the peak value of the projected blood flow speed in the right blood vessel is around 3 mm/s. However, when the phase noise is not well suppressed in the case of Fig. 9(b), the maximum blood flow speed becomes only ~2 mm/s. The complete suppression of phase noise due to motion artifact is important for the accurate evaluation of ODT information.
The comparisons of ODT results between different selected depth ranges, similar to those in Figs. 8(b)-8(d), corresponding to the OCT scanning result in Fig. 2(a) are demonstrated in Figs. 10(b) -10(d). Again, the comparisons show that the selection of a large depth range for searching the phase shift value of maximum counts is useful for suppressing the phase noise due to a motion artifact. The line-scan profiles of projected blood flow speed along the horizontal dashed lines in Figs. 10(b)-10(d) are plotted to give the spiky (red) curves (labeled by “Data”) in Figs. 11(a)-11(c), respectively. Similar to those in Figs. 9(a)-9(c), a smoother curve (labeled by “Fitting”) is plotted after the high spatial-frequency components are filtered in each of Figs. 11(a)-11(c). Here, one can again see that when the phase noise is not completely suppressed, the evaluated flow speed is reduced.
The precise determination of blood vessel locations in advance can effectively help in ODT image analysis when a blood vessel of a large cross section exists in an OCT image. In Figs. 12(a) -12(d), we show the similar results to Figs. 2(a), 2(b), 7(a), and 7(b) of OCT scanning on human skin with the scanning probe. In Fig. 12(c), one can see a large blood vessel cluster at the lower-left corner. Such a blood vessel distribution of a large cross section may disturb the search for the phase shift of maximum counts. The two dashed squares in Fig. 13(a) represent two different depth ranges for searching the phase shifts of maximum counts. When we purposely avoid the depth range containing the blood vessels (range I), we can obtain a clear ODT image as shown in Fig. 13(b). If we do not know the locations of blood vessels in advance and make the depth range selection as II, we would have the ODT image shown in Fig. 13(c), in which the ODT image quality is degraded. Therefore, it is important to first know the precise locations of blood vessels for obtaining high-quality ODT images, particularly when a blood vessel cluster of a large cross section may exist in the image. To further illustrate the difference between the results of Figs. 13(b) and 13(c), we again plot the line-scan profiles of projected blood flow speed along the two upper dashed lines in Figs. 13(b) and 13(c) to give Figs. 14(a) and 14(b), respectively, and along the two lower dashed lines to give Figs. 14(c) and 14(d), respectively. Several arrows are drawn to indicate the locations of blood vessels. Also, smooth curves are plotted after high spatial frequency filtering for clearly demonstrating the flow speeds. Here, again without the complete suppression of phase noise, the evaluated flow speeds are reduced, as shown in the cases of Figs. 14(b) and 14(d).
To show that the condition of large blood vessel cross section in Fig. 12 is often encountered, in Figs. 15 -17 , we show another case of human skin scan with the probe-connected OCT system. Figures 15-17 are similar to Figs. 12-14, respectively. As shown in Fig. 15(c), a blood vessel cluster can be seen in the lower-right corner of the image. With the large cross section of blood vessel distribution, if the depth range selected for searching the maximum-count phase shift covers the blood vessel cluster (range II in Fig. 16(a) ), the phase noise suppression becomes quite poor, as shown in Fig. 16(c). In this situation, the blood flow speed can be underestimated, as demonstrated in Fig. 17(b), when compared with Fig. 17(a).
In summary, we have demonstrated a different image processing procedure for suppressing the phase noise due to a motion artifact acquired during OCT scanning and effectively illustrating the blood vessel distribution in a living tissue. Although both our processing procedure and the widely used procedure (the reference procedure) for micro-angiography application were based on the same concept of selecting the high-frequency components in the spatial-frequency spectrum of B-mode scanning (x-space), by switching the processing order between the x- and k-space, our processing procedure could effectively suppress the phase noise due to a motion artifact. After the blood vessel positions were accurately determined, high-quality ODT images could be obtained with a more careful calibration based on a previously reported method. Our new procedure is useful for clinical micro-angiography application, in which a stepping motor of generating motion artifacts is usually used in the scanning probe.
This research was supported by National Science Council and National Health Research Institute, The Republic of China, under the grants of NSC 99-2218-E-002-013, NSC 99-3114-B-002-005, and NHRI-EX100-10043EI.
References and links
3. Z. Chen, T. E. Milner, S. Srinivas, X. Wang, A. Malekafzali, M.J.C. van Gemert, and J. S. Nelson, “Noninvasive imaging of in vivo blood flow velocity using optical Doppler tomography,” Opt. Lett. 22(14), 1119–1121 (1997). [CrossRef] [PubMed]
4. Y. Zhao, Z. P. Chen, C. Saxer, S. Xiang, J. F. de Boer, and J. S. Nelson, “Phase-resolved optical coherence tomography and optical Doppler tomography for imaging blood flow in human skin with fast scanning speed and high velocity sensitivity,” Opt. Lett. 25(2), 114–116 (2000). [CrossRef] [PubMed]
5. J. A. Izatt, M. D. Kulkarni, S. Yazdanfar, J. K. Barton, and A. J. Welch, “In vivo bidirectional color Doppler flow imaging of picoliter blood volumes using optical coherence tomography,” Opt. Lett. 22(18), 1439–1441 (1997). [CrossRef] [PubMed]
6. A. Szkulmowska, M. Szkulmowski, D. Szlag, A. Kowalczyk, and M. Wojtkowski, “Three-dimensional quantitative imaging of retinal and choroidal blood flow velocity using joint Spectral and Time domain Optical Coherence Tomography,” Opt. Express 17(13), 10584–10598 (2009). [CrossRef] [PubMed]
7. B. Baumann, B. Potsaid, M. F. Kraus, J. J. Liu, D. Huang, J. Hornegger, A. E. Cable, J. S. Duker, and J. G. Fujimoto, “Total retinal blood flow measurement with ultrahigh speed swept source/Fourier domain OCT,” Biomed. Opt. Express 2(6), 1539–1552 (2011). [CrossRef] [PubMed]
8. Z. Zhi, W. Cepurna, E. Johnson, T. Shen, J. Morrison, and R. K. Wang, “Volumetric and quantitative imaging of retinal blood flow in rats with optical microangiography,” Biomed. Opt. Express 2(3), 579–591 (2011). [CrossRef] [PubMed]
9. F. Jaillon, S. Makita, E. J. Min, B. H. Lee, and Y. Yasuno, “Enhanced imaging of choroidal vasculature by high-penetration and dual-velocity optical coherence angiography,” Biomed. Opt. Express 2(5), 1147–1158 (2011). [CrossRef] [PubMed]
10. B. A. Standish, K. K. C. Lee, X. Jin, A. Mariampillai, N. R. Munce, M. F. G. Wood, B. C. Wilson, I. A. Vitkin, and V. X. D. Yang, “Interstitial Doppler optical coherence tomography as a local tumor necrosis predictor in photodynamic therapy of prostatic carcinoma: an in vivo study,” Cancer Res. 68(23), 9987–9995 (2008). [CrossRef] [PubMed]
11. Y. Wang, A. Fawzi, O. Tan, J. Gil-Flamer, and D. Huang, “Retinal blood flow detection in diabetic patients by Doppler Fourier domain optical coherence tomography,” Opt. Express 17(5), 4061–4073 (2009). [CrossRef] [PubMed]
12. S. Makita, Y. Hong, M. Yamanari, T. Yatagai, and Y. Yasuno, “Optical coherence angiography,” Opt. Express 14(17), 7821–7840 (2006). [PubMed]
14. L. An and R. K. Wang, “In vivo volumetric imaging of vascular perfusion within human retina and choroids with optical micro-angiography,” Opt. Express 16(15), 11438–11452 (2008). [CrossRef] [PubMed]
16. L. An, H. M. Subhush, D. J. Wilson, and R. K. Wang, “High-resolution wide-field imaging of retinal and choroidal blood perfusion with optical microangiography,” J. Biomed. Opt. 15(2), 026011 (2010). [CrossRef] [PubMed]
17. Z. Zhi, Y. Jung, Y. Jia, L. An, and R. K. Wang, “Highly sensitive imaging of renal microcirculation in vivo using ultrahigh sensitive optical microangiography,” Biomed. Opt. Express 2(5), 1059–1068 (2011). [CrossRef] [PubMed]
19. M. T. Tsai, C. K. Lee, H. C. Lee, H. M. Chen, C. P. Chiang, Y. M. Wang, and C. C. Yang, “Differentiating oral lesions in different carcinogenesis stages with optical coherence tomography,” J. Biomed. Opt. 14(4), 044028 (2009). [CrossRef] [PubMed]
20. C. C. Yang, M.-T. Tsai, H.-C. Lee, C.-K. Lee, C.-H. Yu, H.-M. Chen, C.-P. Chiang, C.-C. Chang, Y.-M. Wang, and C. C. Yang, “Effective indicators for diagnosis of oral cancer using optical coherence tomography,” Opt. Express 16(20), 15847–15862 (2008). [CrossRef] [PubMed]
21. R. K. Manapuram, V. G. R. Manne, and K. V. Larin, “Development of phase-stabilized swept-source OCT for the ultrasensitive quantification of microbubbles,” Laser Phys. 18(9), 1080–1086 (2008). [CrossRef]
22. J. Sun, S. Guo, L. Wu, L. Liu, S. W. Choe, B. S. Sorg, and H. Xie, “3D in vivo optical coherence tomography based on a low-voltage, large-scan-range 2D MEMS mirror,” Opt. Express 18(12), 12065–12075 (2010). [CrossRef] [PubMed]
23. K. H. Kim, B. H. Park, G. N. Maguluri, T. W. Lee, F. J. Rogomentich, M. G. Bancu, B. E. Bouma, J. F. de Boer, and J. J. Bernstein, “Two-axis magnetically-driven MEMS scanning catheter for endoscopic high-speed optical coherence tomography,” Opt. Express 15(26), 18130–18140 (2007). [CrossRef] [PubMed]