Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

Speckle reduction in visible-light optical coherence tomography using scan modulation

Speckle reduction in visible-light optical coherence tomography using scan modulation Speckle reduction in visible-light optical coherence tomography using scan modulation Ian Rubinoff Lisa Beckmann Yuanbo Wang Amani A. Fawzi Xiaorong Liu Jenna Tauber Katie Jones Hiroshi Ishikawa Joel S. Schuman Roman Kuranov Hao F. Zhang Ian Rubinoff, Lisa Beckmann, Yuanbo Wang, Amani A. Fawzi, Xiaorong Liu, Jenna Tauber, Katie Jones, Hiroshi Ishikawa, Joel S. Schuman, Roman Kuranov, Hao F. Zhang, “Speckle reduction in visible-light optical coherence tomography using scan modulation,” Neurophoton. 6(4), 041107 (2019), doi: 10.1117/1.NPh.6.4.041107. Neurophotonics 6(4), 041107 (Oct–Dec 2019) Speckle reduction in visible-light optical coherence tomography using scan modulation a a b c d e e Ian Rubinoff, Lisa Beckmann, Yuanbo Wang, Amani A. Fawzi, Xiaorong Liu, Jenna Tauber, Katie Jones, e e a,b a,c, Hiroshi Ishikawa, Joel S. Schuman, Roman Kuranov, and Hao F. Zhang * Northwestern University, Department of Biomedical Engineering, Evanston, Illinois, United States Opticent Health, Evanston, Illinois, United States Northwestern University, Department of Ophthalmology, Chicago, Illinois, United States University of Virginia, Department of Biology and Psychology, Charlottesville, Virginia, United States New York University, Department of Ophthalmology, New York, United States Abstract. We present a technique to reduce speckle in visible-light optical coherence tomography (vis-OCT) that preserves fine structural details and is robust against sample motion. Specifically, we locally modulate B-scans orthogonally to their axis of acquisition. Such modulation enables acquisition of uncorrelated speckle patterns from similar anatomical locations, which can be averaged to reduce speckle. To verify the effectiveness of speckle reduction, we performed in-vivo retinal imaging using modulated raster and circular scans in both mice and humans. We compared speckle-reduced vis-OCT images with the images acquired with unmodulated B-scans from the same anatomical locations. We compared contrast-to-noise ratio (CNR) and equivalent num- ber of looks (ENL) to quantify the image quality enhancement. Speckle-reduced images showed up to a 2.35-dB improvement in CNR and up to a 3.1-fold improvement in ENL with more discernable anatomical features using eight modulated A-line averages at a 25-kHz A-line rate. © The Authors. Published by SPIE under a Creative Commons Attribution 4.0 Unported License. Distribution or reproduction of this work in whole or in part requires full attribution of the original publication, including its DOI. [DOI: 10.1117/1.NPh.6.4.041107] Keywords: optical coherence tomography; visible light; speckle; retina; imaging; clinic. Paper 19045SSR received May 13, 2019; accepted for publication Aug. 8, 2019; published online Sep. 3, 2019. uncorrelated speckle patterns. Uncorrelated patterns from sim- 1 Introduction ilar structural locations can be averaged to remove the speckle Optical coherence tomography (OCT) is a scattering-based and reveal the original anatomical information. The physical imaging technology that acquires high-resolution three- 1 basis of incoherent averaging method makes it ideal for situa- dimensional images of biological samples in vivo. Following tions where the study of fine anatomical features is required. its initial report in 1991, OCT has become the “gold standard” However, manipulating image acquisition to obtain uncorrelated for noninvasive retinal imaging. Today, it is an essential technol- speckle patterns can be challenging. First, different scattering ogy in labs and clinics for studying and managing a wide variety 2 events must be probed without losing the structural integrity of retinal diseases. Advances in optoelectronics in the past of the location of interest. Second, multiple acquisitions at a par- 25 years has led to improved resolution, signal-to-noise ratio ticular location are required to generate enough patterns suitable (SNR), and imaging field of view (FOV) in OCT. However, for averaging. Samples with strong motion can pose a challenge speckle, an image artifact caused by the self-interference of to averaging, especially in human eyes. Multiple approaches coherent light at random phases, remains a significant source 4 have been developed to achieve incoherent averaging of speckle of reduced image quality. This is of particular salience in retinal while retaining high image quality. The most basic technique is imaging, where speckle noise can obscure fine structures in the to average consecutive B-scans in a raster pattern, either from outer retina, such as the retinal pigment epithelium (RPE) and the same location or from a slightly offset position. The former Bruch’s membrane (BM). Minute pathological changes in these relies on a small sample movement to modify scattering events, structures may be strongly associated with the progressions of and the latter directly modifies scattering events across consecu- several retinal diseases, including macular degeneration and tive separated B-scans. More advanced techniques for spatial central serous retinopathy (CSR). averaging include modulating the scanning beam after every To improve the imaging quality, researchers have developed 10 11 A-line with a translational offset or angular offset. In particu- several methods to suppress speckle artifacts in OCT. These lar, translational offset has been shown to be more robust against methods can be classified into two categories: digital filtering sample motion than B-scan averaging. Both scan modulations and incoherent averaging. Digital filtering, while simple to have been previously implemented with additional hardware implement and effective in reducing the grainy appearance of and moving parts. Other techniques include multiwavelength speckle, causes blurring that degrades image resolution and pre- 12,13 14 averaging, modulation of light wavefront, and nonlocal, vents the delineation of fine anatomical features. Incoherent software-based averaging. averaging, on the other hand, samples photons that have under- Recent development of visible-light optical coherence gone statistically different scattering events, thereby generating tomography (vis-OCT) has generated new capabilities for retinal imaging, including visualization of fine structures with ultrahigh resolution and spectroscopic analysis of blood-oxygen concen- *Address all correspondence to Hao F. Zhang, E-mail: hfzhang@north 16–18 western.edu tration (sO ). Speckle, which distorts both structural and Neurophotonics 041107-1 Oct–Dec 2019 Vol. 6(4) Rubinoff et al.: Speckle reduction in visible-light optical coherence. . . spectral information, can undermine these new benefits. To reduce speckle in vis-OCT using incoherent averaging, there are some unique challenges. First, there is strong optical absorption and scattering in tissue in the visible-light spectral range, reduc- ing the amount of photons that can be collected per unit time. This is coupled with high relative intensity from the super- continuum laser source. To achieve high SNR, a prolonged camera exposure is required, therefore reducing scanning speed. Second, since eyes are sensitive to visible-light illumination, vis-OCT often suffers from increased retinal motion. Finally, preservation of structure-dependent spectroscopic information requires anatomically localized sampling of scattering events. A method for speckle reduction that is both structurally local- ized and robust against retinal motion is optimal for vis-OCT. Furthermore, it is ideal for a speckle reduction technique to be easily implementable in clinical settings to increase usability, reduce likelihood of malfunction, and cause no additional discomfort to patients. A straightforward approach is to perform B-scan averaging. However, eye motion in vis-OCT makes B-scan averaging unreliable due to blurring, even with post- acquisition image registration. To overcome this challenge, we modulated the scanning trajectory orthogonally to the direction of the B-scan images during data acquisition. This method Fig. 1 Illustrations of speckle-reduction scanning protocols. enabled sampling of different speckle patterns while maintain- (a) Overall illustration of the relationship between the B-scan axis and the orthogonal axis in the modulated raster scan; (b) detailed illustra- ing high anatomical similarity between modulations. We imple- tion of the A-line acquisition sequence in the modulated raster scan. mented scanning modulation by directly controlling the Here, d is the distance between two adjacent A-lines along the galvanometer scanners without additional hardware. orthogonal axis. The arrows 1, 2, and 3 highlight the trajectory of galvanometer motion. (c) Overall illustration of the relationship be- tween the B-scan axis and the orthogonal axis in the modulated cir- 2 Methods cular scan; (d) detailed illustration of the A-line acquisition sequence in the modulated circular scan. 2.1 Scanning Protocol We modulated both raster and circular scans, which are com- We averaged all n A-lines in the orthogonal direction along monly used in vis-OCT, to test our speckle reduction method. each rectangular edge [Figs. 1(b) and 1(d)] to generate a single Figure 1 illustrates the modulated raster scan [Figs. 1(a) and speckle-reduced A-line (srA-line). For a desired sampling den- 1(b)] and modulated circular scan [Figs. 1(c) and 1(d)]. As sity of m srA-lines per speckle-reduced B-scan (srB-scan), the shown in Figs. 1(a) and 1(c), we define the B-scan axis as the total number of camera acquisitions per srB-scan is n × m. Each direction along which a traditional cross-sectional image would consecutive srA-line in an srB-scan can then be calculated as be acquired without modulation. We define the orthogonal axis as the direction orthogonal to the B-scan axis on the two- dimensional scanning plane. Movement along the orthogonal EQ-TARGET;temp:intralink-;e001;326;328srA ¼ A for j ¼ 0; 1; 2;:::m − 1; (1) j ij axis [arrow 1 in Figs. 1(b) and 1(d)] occurs in n equidistant i¼1 steps, where n is the number of speckle-uncorrelated A-lines where i is the index of each set of n A-lines about the edge of a to be locally averaged. Each translation of the galvanometer [red dots in Figs. 1(b) and 1(d)] is discrete, synchronized with the rectangle; j is the index of each consecutive edge of a rectangle; spectrometer camera exposure, and implemented entirely via and A is the i þ j × n’th A-line in a full B-scan acquisition. ij software control without additional hardware or moving parts. Since an srB-scan increases imaging time over normal This avoids a complex synchronization procedure or risk of B-scan acquisition by a factor of n, it is important to collectively desynchronization between the beam path and the camera expo- limit n, m, and the number of total srB-scans to prevent overly sure when using an external scanner. The centroids of each long imaging time. First, all sampling numbers were selected spot generating an A-line are separated by a distance d along in powers of 2 to support fast graphics processing unit data the orthogonal axis [Fig. 1(b)]. After n translations in this direc- processing. Next, we limited all imaging experiments to 8192 tion, the beam is shifted along the B-scan axis [arrow 2 in total A-lines per srB-scan. Given a camera exposure time of Figs. 1(b) and 1(d)], followed by a reversed scan along the 40 μs, which is required for sufficiently high SNR, an srB-scan orthogonal axis [arrow 3 in Figs. 1(b) and 1(d)]. Such modula- could be acquired in 328 ms, an upper limit for reducing bulk tion superimposes a rectangular wave on the B-scan axis, where motion artifacts (satisfying Nyquist criterion of 500 ms for eye each rising and falling edge of each rectangle contains n microsaccades of ∼1Hz). Furthermore, we chose to limit the speckle-uncorrelated A-lines. While other modulation shapes total image acquisition time to ∼5s to prevent patient fatigue such as sinusoidal or triangular are possible, we chose rectan- and discomfort. This limited the total amount of srB-scans per gular to best preserve lateral resolution along the B-scan axis. acquisition to 16 (5.25 s total acquisition time). In our experi- During acquisition, several parameters, including n, d, and mental human imaging system (Sec. 2.3.2), we maximized imaging FOV are adjustable. We investigate how to obtain an lateral sampling density in a raster scan without spot overlap, optimal d in Sec. 3.1. where m ¼ 1024 srA-lines and n ¼ 8 averages. The parameters Neurophotonics 041107-2 Oct–Dec 2019 Vol. 6(4) Rubinoff et al.: Speckle reduction in visible-light optical coherence. . . n, m, and d can be easily modified for different experimental between the reference B-scan and srB-scans to evaluate the conditions. effectiveness of speckle reduction. The srB-scan averages along the same locations as n spatially separated B-scans, each of m A-lines. However, in direct B-scan 2.3 Data Acquisition averaging, each A-line at a particular lateral position is delayed temporally by the scanner’s fly-back time. This results in a total We tested our speckle reduction protocol in both mouse and sampling period of n × m × t, where t is the camera exposure human retinas using two prototype systems developed at time for each A-line. In our method, modulation removes the Northwestern University. In addition, we further tested our wait for scanner fly-back, thereby reducing the total sampling speckle reduction method in humans in a clinical setting using period to n × t for each srA-line. Using our experimental param- a commercial vis-OCT system (Aurora X1, Opticent Health), eters (m ¼ 1024, n ¼ 8, t ¼ 40 μs), we reduced the sampling where optical engineering expertise was unavailable. We period from 328 ms to 320 μs for each srA-line and increased directly implemented the modulated scanning protocol in that the srA-line rate from 3 to 3125 Hz. Constant, involuntary reti- system without additional calibration, alignment, or changes nal motions can occur at frequencies up to 90 Hz with ampli- to the photographer’s workflow. tudes up to 40 arc sec (equivalent to 0.011-μm change in sampling location per 40-μs camera exposure in the human 2.3.1 Mouse imaging retina). This leaves the possibility of only 0.088 μm of move- ment during an srA-line, which is insignificant when compared For mouse imaging, we used the system described in our pre- 22 2 with the micron-order lateral and axial resolutions in OCT. vious work. In brief, a 1∕e spot size of ∼5.7 μm was incident Therefore, the improved srA-line rate is highly significant. on the retina. We controlled the total illumination power to 1.2 mW on the cornea in all instances. For a raster scan, we used m ¼ 1024, n ¼ 8, and d ¼ 6.3 μm.AnFOVof 1.4 × 1.4 mm 2.2 Metrics to Evaluate Image Quality Improvement was used in all mouse retina images, equivalent to ∼1.4-μm We used contrast-to-noise ratio (CNR) and equivalent number separation between srA-lines along the B-scan axis. For a cir- of looks (ENL) to evaluate image quality improvement after cular scan, we used n ¼ 8 and d ¼ 5.4 μm. The circle circum- speckle reduction. CNR measures how well the sample feature ference was 1.8 mm, equivalent to ∼1.8-μm separation between can be discerned from the surrounding background. Mean inten- srA-lines along the B-scan axis. Improved from our previous sity and variance from both the image background and the system, we adopted a commercial spectrometer (Blizzard SR, sample feature are included to account for two separate noise Opticent Health) with a 2048-pixel line scan camera covering components: intrinsic OCT background noise and speckle. 510 to 610 nm, which provided an axial resolution of ∼1 μm Since the optical properties of different features vary, we calcu- in tissue. We used an A-line rate of 25 kHz in all rodent lated the CNR (dB) in confined region of interests (ROIs) as experiments. All rodent experimental procedures were approved by the μ − μ i b Northwestern University IACUC and conformed to the Asso- EQ-TARGET;temp:intralink-;e002;63;403CNR ¼ 10 log pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ; (2) 2 2 σ þ σ ciation for Research in Vision and Ophthalmology (ARVO) i b Statement on Animal Research. We anesthetized adult where μ is the mean intensity of the i’th ROI; μ is the mean i b C57Bl6/J mice (n ¼ 8) with an intraperitoneal injection intensity of the background outside of the sample structure; σ is (10 mL∕kg body weight) of a ketamine/xylazine cocktail the variance of the i’th ROI; and σ is the variance of the back- (ketamine: 11.45 mg∕mL; xylazine: 1.7 mg∕mL). Each mouse ground outside of the sample feature. was then placed on a custom-made animal holder and immobi- ENL is the squared inverse of the speckle contrast and mea- lized for imaging. The body temperature was maintained with sures the smoothness and homogeneity within an ROI. We cal- a heat lamp. To dilate the pupil, we applied a drop of 1% culated ENL as tropicamide hydrochloride ophthalmic solution. Throughout imaging, we applied one drop of commercial artificial tears after each image acquisition to prevent corneal dehydration. EQ-TARGET;temp:intralink-;e003;63;283ENL ¼ ; (3) After the imaging session concluded, the mouse was allowed to recover under heat lamp and was returned to the animal where μ is the mean intensity of the i’th ROI and σ is the vari- housing facility. ance of the i’th ROI. An increase in ENL serves as a strong indicator for the reduction of speckle. 2.3.2 Human imaging We compared CNR and ENL in srB-scans with a “reference” B-scan from the same location as an srB-scan. A reference Human imaging was performed using two vis-OCT systems. B-scan included 8192 A-lines acquired along the B-scan axis First, images were acquired in the Ophthalmology Department with a 40-μs camera exposure. Every eight consecutive A-lines at Northwestern Memorial Hospital using an experimental sys- were averaged, resulting in a final sampling density of 1024 tem reported in our previous work. We controlled the illumi- averaged A-lines per reference B-scan. This operation was nation power to be <250 μW on the cornea in all our human equivalent to acquiring an srB-scan without modulating the studies. A 1∕e spot size of ∼6.3 μm was incident on the retina. scanner along the orthogonal axis. Because of high sampling For a raster scan, we used n ¼ 8, d ¼ 7 μm, and m ¼ 1024. The density along the B-scan axis, averaged speckle patterns were FOV was 6.8 × 6.8 mm , leading to ∼6.6-μm separation be- still highly correlated, preventing reduction of speckle. tween srA-lines along the B-scan axis. For a circular scan, However, background noise was equally suppressed in reference we used n ¼ 8 and d ¼ 5.9 μm. The circle circumference was B-scans and srB-scans because they used the same amount of 18.3 mm, equivalent to ∼18-μm separation between srA-lines temporal averaging. We compared the CNR and ENL values along the B-scan axis. Similar to our mouse system, we Neurophotonics 041107-3 Oct–Dec 2019 Vol. 6(4) Rubinoff et al.: Speckle reduction in visible-light optical coherence. . . upgraded to a commercial spectrometer (Blizzard SR, Opticent Figures 2(g)–2(j) show the similar comparison in the circular Health). The A-line rate was 25 kHz in all human imaging scan, where the speckle-reduced circular scan with d ¼ 6 μm tests. demonstrates improvement in image quality. To verify the efficacy of our method outside the lab environ- Figures 2(k) and 2(l), respectively, show the pixel intensity ment, we further conducted human imaging in the Department histograms from the tape layer 1 [highlighted in Fig. 2(c)] in the of Ophthalmology at New York University (NYU) Langone raster and circular scans. In both scan patterns, the intensity his- Medical Center. A clinical vis-OCT system (Aurora X1, tograms changed from a broad, right-skewed distribution when Opticent Health) was used to acquire all images. It offered d ¼ 0 μm to a lower-variance, nearly centrosymmetric distribu- an axial resolution of ∼1 μm and we controlled the spectrometer tion when d ¼ 6 μm. These results agree with the expected exposure time to be 40 μs. We used the same raster scanning change in pixel intensity distribution from Rayleigh distribution parameters as those in the Northwestern system but reduced the to Poisson distribution after speckle reduction. FOV to 5 × 5mm . We implemented the speckle-reduction As shown in Figs. 2(a) and 2(b), we used a −0.25-dB drop acquisitions in Aurora X1 entirely via a software update without in CNR to determine the range of acceptable d values, which any additional calibration or hardware modifications. Clinical gives d ¼ 5.1 μm and d ¼ 7.8 μm. This range is helpful min max photographers acquired retinal images without any changes to for human retinal imaging, where the eye shape, optical proper- their normal workflow. ties, and scanning location may differ among subjects. We also All human imaging procedures in the respective imaging noted that the optimal d ¼ 6 μm is approximately equal to the locations were approved by the Northwestern University estimated spot size of ∼5.5 μm on the retina. This suggests Institutional Review Board (IRB) and NYU IRB and adhered that adjacent spots along the orthogonal axis should be as close to the tenets of the Declaration of Helsinki. Healthy volunteers as possible without spatial overlap. This result is consistent with without known eye diseases provided informed consent before the notion that spatial overlapping provides correlated speckle imaging (Northwestern site: n ¼ 3; NYU site: n ¼ 6). patterns. This result also suggests that it is acceptable to estimate the optimal d using the OCT focal spot size on the sample. These considerations are not expected to change in the living 2.4 Initial Calibration for Orthogonal Spot human eye, where local movement during a single srA-line Separation (0.088 μm) is significantly less than d. We adjusted the d value within the identified range in rodent A calibration procedure was needed for coarse determination and human imaging to accommodate different eye conditions. of optimal spot separation, d, along the orthogonal axis. Since Since we control the d value by the galvanometer angle, we CNR is associated with the ability to discern features from noise, identified optimal angular step size along the orthogonal axis we used it as the primary indicator for image quality. In theory, in different experimental conditions. For mouse imaging, the an increased d increases the decorrelation of the speckle patterns optimal angular step sizes were 0.175 deg and 0.15 deg, which between adjacent orthogonal A-lines. After averaging, speckle correspond to d values of 6.3 and 5.4 μm, respectively, in raster is maximally reduced when the averaged patterns are entirely and circular scans. For human imaging, the optimal angular step uncorrelated. However, if d is too large, we will lose structural sizes were 0.025 deg and 0.02 deg, which correspond to d values similarity between orthogonal A-lines, which can result in of 7 and 5.9 μm, respectively, in raster and circular scans. image blurring. We investigated the impact of modulation dis- tance on CNR by imaging a model mouse eye using both raster and circular scans. The model eye was made from a silica bead (diameter: 3.15 mm). We attached two layers of tape and paper 3.2 Speckle Reduction in the Mouse Retina with an ink pattern to the bottom of the bead to simulate the Figure 3 shows the speckle reduction results in a mouse retina retinal layers. Using the rodent vis-OCT system, we reached using a raster scan. Figures 3(a) and 3(b) are the reference a 1∕e spot size of ∼5.5 μm on the tape layers through the bead. B-scan and srB-scan images, respectively. The imaged retina We then varied the d value from 0 to 13.75 μm in 16 steps and in the srB-scan has a smoother, less grainy appearance that pro- acquired an srB-scan after each step. We calculated CNR from vides a clearer differentiation between anatomical layers. We three ROIs in the top tape layer and averaged them to determine selected six ROIs from the inner plexiform layer [IPL, high- the impact of the d value on image quality. lighted by (a1) and (b1)], outer nuclear layer [ONL, highlighted by (a2) and (b2)], and outer retinal layer [ORL, highlighted by 3 Results (a3) and (b3)] to quantify quality improvement. Figures 3(a1)– 3(a3) and Figs. 3(b1)–3(b3) show the magnified views of the six 3.1 Impact of Modulation Distance on Image Quality selected ROIs and Table 1 shows the quantitative comparisons The results to identify an optimal d value are shown in Fig. 2. of CNR and ENL values from these ROIs. Speckle reduction is Figures 2(a) and 2(b), respectively, show how CNR values vary particularly helpful in the ORL, where a small gap near RPE and as a function of d in imaging the model eye using modulated BM layers is revealed [Fig. 3(b3)], which is not visible in the raster and circular scans. When d is increased from 0 to reference B-scan image [Fig. 3(a3)]. The capability to differen- 13.75 μm in both scans, CNR reaches its maximum at d ¼ tiate RPE and BM may add significant value to various preclini- 6 μm. Figures 2(c) and 2(e) show raster srB-scans with d ¼ 0 cal studies using mouse models. and 6 μm, respectively. Figures 2(d) and 2(f) show magnified Figure 4 shows the speckle reduction results in a mouse views of the two highlighted images [yellow boxes in Figs. 2(c) retina using circular scan. Figures 4(a) and 4(b) are the reference and 2(e)], respectively. The srB-scan with d ¼ 6 μm [Fig. 2(f)] B-scan and srB-scan images, respectively. Again, the srB-scan shows a smoother intensity distribution within each layer and image improved the overall image quality with better differen- much improved discrimination between the tape and the paper tiated fine anatomical features. We also selected six ROIs from layers, as compared with the unmodulated scan [Fig. 2(d)]. IPL [highlighted by (a1) and (b1)] and retinal blood vessels Neurophotonics 041107-4 Oct–Dec 2019 Vol. 6(4) Rubinoff et al.: Speckle reduction in visible-light optical coherence. . . Fig. 2 Speckle-reduction test in the model mouse eye. (a) Change of averaged CNR as a function of d in the modulated raster scan; (b) change of averaged CNR as a function of d in the modulated circular scan. (c) An srB-scan image of the model mouse eye acquired using modulated raster scan with d ¼ 0 μm. The structures corresponding to the two tape and one paper layers are highlighted by the arrows. (d) Magnified view of the region highlighted by the box in panel (c). (e) An srB-scan image of the model mouse eye acquired using modulated raster scan with d ¼ 6 μm. (f) Magnified view of the region high- lighted by the box in panel (e). (g) An srB-scan image of the model mouse eye acquired using modulated circular scan with d ¼ 0 μm. (h) Magnified view of the region highlighted by the box in panel (g). (i) An srB-scan image of the model mouse eye acquired using modulated circular scan with d ¼ 6μm. (j) Magnified view of the region highlighted by the box in panel (i). All images are plotted with identical color bar. (k) Fitted pixel-intensity histograms within the tape layer 1 acquired by modulated raster scans with d ¼ 0 and 6 μm; (l) fitted pixel-intensity histograms within the tape layer 1 acquired by modulated circular scans with d ¼ 0 and 6 μm. [highlighted by (a2), (b2), (a3), and (b3), respectively] for quan- For circular scans, the ROIs in the IPL and two vessels show titative evaluation. 1.84, 1.90, and 1.11 dB respective improvement in CNR, and Table 1 compares the CNR and ENL values from the selected 2.69, 2.56, and 1.63 times respective improvements in ENL. ROIs in both raster and circular scans. In each scan mode, we CNR and ENL improvements for the ORL in the raster scan and see increased metric values from the ROIs in the srB-scan second vessel in the circular scan are slightly lower than other images. For raster scans, the ROIs in the IPL, ONL, and ORL improvements. This is because some of the image background is show 2.35, 1.84, and 1.32 dB respective improvements in CNR, unavoidably included in the ROI, artificially contributing low and 3.1, 2.53, and 1.84 times respective improvements in ENL. pixel intensities to μ in the metrics. Neurophotonics 041107-5 Oct–Dec 2019 Vol. 6(4) Rubinoff et al.: Speckle reduction in visible-light optical coherence. . . Fig. 3 Speckle-reduction test in mouse retina using modulated raster scan. (a) Reference raster B-scan image. Three ROIs from IPL, ONL, and ORL are highlighted by (a1), (a2), and (a3), respectively. The size of each ROI is 140 μmðlateralÞ × 20 μmðaxialÞ. (b) The corresponding srB-scan image. The same three ROIs are highlighted by (b1), (b2), and (b3), respectively. (a1)–(a3) The magnified views of the three highlighted ROIs in panel (a). (b1)–(b3) The magnified views of the three highlighted ROIs in panel (b). CNR and ENL values are calculated from all the selected ROIs. All images are plotted with identical color bar. 3.3 Speckle Reduction in the Human Retina RPE, and BM become clearly discernable from one another. The thickness of BM is measured as ∼3 μm and is resolved in the We accomplished speckle reduction in the human retina using whole image without blur or distortion. The average measured both the laboratorial prototype and a clinical vis-OCT system. 23 thickness of BM in the human eye is ∼2 to 5 μm, which is Unlike mouse imaging, in which retinal motion can be mini- consistent with our measurement. The distinct separation mized and images can be acquired over an extended period, between the BM and the RPE, as shown in Fig. 5(a3), may open human imaging usually suffers from severe retinal motions and up new window to investigate macular degeneration, where image acquisition needs to complete within few seconds. For vis- initial pathological alterations are hypothesized to start from 23,24 OCT, retinal motion can be much stronger as described in Sec. 1. BM. Finally, we note a shadow caused by a small blood ves- Figure 5 shows the speckle reduction results using raster scan sel as highlighted by the arrows in both the reference B-scan in a human retina (22-year-old male volunteer). Figures 5(a) and [Fig. 5(a3)] and the srB-scan [Fig. 5(b3)] images in ORL. It 5(b) are reference B-scan and srB-scan images superior to the is measured as 2 pixels laterally or ∼14 μm in width. This fea- optic disk, respectively. The srB-scan is smoother and less ture is better resolved in the srB-scan image, indicating that lat- grainy in appearance than the reference B-scan, increasing vis- eral resolution has been well preserved after speckle reduction. ibility of the retinal layers. Improved image quality here is con- Repetitive B-scan averaging is not trivial due to retinal sistent with that in the mouse retina [Fig. 3(b)]. We selected six motion, which often leads to image blurring even after registra- ROIs from the nerve fiber layer [NFL, highlighted by (a1) and tion. We overcame this challenge and showed that our speckle (b1)], ganglion cell layer [GCL, highlighted by (a2) and (b2)], reduction method is robust against retinal motion in Fig. 6. and ORLs [highlighted by (a3) and (b3)] to quantify quality We acquired eight repeated raster B-scans (each containing improvement. Figures 5(a1)–5(a3) and 5(b1)–5(b3) show the 1024 A-lines) from the same anatomical location and volunteer, magnified views of the six selected ROIs and Table 2 shows the as shown in Figs. 5(a) and 5(b). All the B-scans were axially quantitative comparisons of CNR and ENL values from these and laterally registered using an fast Fourier transform based ROIs. Of particular note is the increased clarity of ORL in the cross-correlation algorithm. The averaged B-scan image srB-scan [Fig. 5(b3)]. Unlike the reference B-scan [Fig. 5(a3)], [Fig. 6(a)] shows blurred anatomical layers in both the inner the shape and boundaries of the rod outer segment tips (ROST), retina and the outer retina due to motion. Figure 6(b) shows Neurophotonics 041107-6 Oct–Dec 2019 Vol. 6(4) Rubinoff et al.: Speckle reduction in visible-light optical coherence. . . Fig. 4 Speckle-reduction test in mouse retina using modulated circular scan. (a) Reference circular B- scan image. Three ROIs from IPL and two vessels are highlighted by (a1), (a2), and (a3), respectively. The size of each ROI is 140 μmðlateralÞ × 20 μmðaxialÞ. (b) The corresponding srB-scan image. The same three ROIs are highlighted by (b1), (b2), and (b3), respectively. CNR and ENL values are calcu- lated from all the selected ROIs. All images are plotted with identical color bar. Table 1 Image quality metric values from the ROIs in the mouse averaged B-scan images acquired by vis-OCT using similar scan retina shown in Figs. 3 and 4. parameters. Figure 6(d) shows a magnified view of the same ana- tomical position from an srB-scan image, where all anatomical layers are clearly resolved across the whole image. The same CNR (dB) CNR (dB) ENL ENL A-line locations from Fig. 6(b) are highlighted in Fig. 6(d) (by Scan type ROI ref. B-scan srB-scan ref. B-scan srB-scan 3 and 4). Figure 6(e) shows A-line 3 and A-line 4, confirming Raster IPL 2.03 4.38 3.94 12.21 that all ORLs are well resolved despite retinal motion. We also demonstrate speckle reduction in circular scan in the Raster ONL 1.63 3.47 3.38 8.56 human retina (Fig. 7). Figures 7(a) and 7(b) show a reference Raster ORL 1.83 3.15 2.76 5.07 B-scan image and an srB-scan image, acquired at the same anatomical location, respectively. We selected six ROIs from Circular IPL 2.32 4.16 4.74 12.74 the same locations as in Fig. 5, including the NFL [highlighted by (a1) and (b1)], GCL [highlighted by (a2) and (b2)], and Circular Vessel 1 1.63 3.53 3.36 8.60 ORL [highlighted by (a3) and (b3)] to quantify quality improve- Circular Vessel 2 2.00 3.11 6.11 9.97 ment. Figures 7(a1)–7(a3) and 7(b1)–7(b3) show magnified views of the six selected ROIs and Table 2 shows the quantita- tive comparisons of the CNR and ENL values from these ROIs. Similar to the ORL in the raster srB-scan [Fig. 6(b3)], the ORL a magnified view of the region highlighted by the box in Fig. 6(a). in the circular srB-scan [Fig. 7(b3)] shows distinct separation Two A-lines from the locations highlighted by lines 1 and 2 in between BM, RPE, and ROST. In the reference B-scan image Fig. 6(b) are shown in Fig. 6(c). A-line 1 reveals five anatomical [Fig. 7(a3)], however, boundaries of these anatomical layers layers in the outer retina, notably with reduced contrast near the are not easily differentiated due to speckles. To the best of our RPE. A-line 2 fails to resolve any anatomical features. Since vis- knowledge, this is the first demonstration of speckle-reduced OCT offers an axial resolution of near 1 μm, small misalignments in B-scan averaging may lead to much severer blurring. The imaging in a circular pattern using localized scan modulation image quality shown in Fig. 6(a) is representative of most in the human retina. Neurophotonics 041107-7 Oct–Dec 2019 Vol. 6(4) Rubinoff et al.: Speckle reduction in visible-light optical coherence. . . Fig. 5 Speckle-reduction test in human retina using modulated raster scan. (a) Reference raster B-scan image. Three ROIs from NFL, GCL, and ORL are highlighted by (a1), (a2), and (a3), respectively. The size of each ROI is 430 μmðlateralÞ × 23 μmðaxialÞ. (b) The corresponding srB-scan image. The same three ROIs are highlighted by (b1), (b2), and (b3), respectively. (a1)–(a3) The magnified views of the three highlighted ROIs in panel (a). (b1)–(b3) The magnified views of the three highlighted ROIs in panel (b). CNR and ENL values are calculated from all the selected ROIs. The bottom three anatomical layers ROST, RPE, and BM are highlighted in panel (b3). The arrows in (a3) and (b3) highlight the same blood vessel shadow. All images are plotted with identical color bar. Fig. 6 Directly comparing averaged B-scan with srB-scan images from human retina. (a) Image scan from the same location as shown in Fig. 5(b) after averaging eight B-scans. (b) Magnified view of the outer retina region as highlighted in panel (a). (c) Two A-lines from the positions highlighted by 1 and 2 in panel (b). (d) Magnified view of the same outer retina region from the srB-scan shown in Fig. 5(b). Five anatomical layers are labeled. (e) Two A-lines from the positions highlighted by 3 and 4 in panel (d). All plotted A-lines are averaged three times laterally to reduce variation. All images are plotted on the same contrast scale as used in Fig. 5. Table 2 compares the CNR and ENL values from the in CNR of 2.08, 1.92, and 1.64 dB, in the NFL, GCL, and selected ROIs in the human retina for both raster and circular ORL, respectively. Here, corresponding ENL improvements scans. In each instance, we see increased metric values in the are 2.77, 2.11, and 1.81 times, respectively. Again, we attrib- srB-scan ROIs. Raster scans show an improvement in CNR of ute the slightly lower metric value increases in the ORL to 2.25,2.00, and1.86dB, in theNFL,GCL,and ORL, respec- the unavoidable inclusion of image background in the ROI, tively. Corresponding ENL improvements are 2.87, 2.72, and artificially contributing low pixel intensities to μ in the 2.52 times, respectively. Circular scans show an improvement metrics. Neurophotonics 041107-8 Oct–Dec 2019 Vol. 6(4) Rubinoff et al.: Speckle reduction in visible-light optical coherence. . . Fig. 7 Speckle-reduction test in human retina using modulated circular scan. (a) Reference circular B- scan image. Three ROIs from NFL, GCL, and ORL are highlighted by (a1), (a2), and (a3), respectively. The size of each ROI is 900 μmðlateralÞ × 40 μmðaxialÞ. (b) The corresponding srB-scan image. The same three ROIs are highlighted by (b1), (b2), and (b3), respectively. (a1)–(a3) The magnified views of the three highlighted ROIs in panel (a). (b1)–(b3) The magnified views of the three highlighted ROIs in panel (b). CNR and ENL values are calculated from all the selected ROIs. The bottom four ana- tomical layers cone outer segment tips (COST), ROST, RPE, and BM are highlighted in panel (b3). All images are plotted with identical color bar. training. Quality improvement in the clinical speckle-reduced Table 2 Image quality metric values from the ROIs in the human images is comparable to the lab tests (Sec. 3.3). Figures 8(a) retina from Figs. 5 and 7. and 8(b), respectively, show a reference B-scan and an srB-scan of the macula from a 37-year-old female volunteer. In the refer- CNR (dB) CNR (db) ENL ENL ence B-scan image, speckle particularly distorts the ORL, as Scan type ROI ref. B-scan srB-scan ref. B-scan srB-scan shown in Fig. 8(c), preventing the delineation of fine anatomical structures, such as the RPE and BM. As a comparison, all five Raster NFL 2.39 4.64 4.19 12.01 ORLs, including the RPE and BM, are clearly resolved in the Raster GCL 2.31 4.31 5.92 16.12 magnified srB-scan image, as shown in Fig. 8(d). Quantitatively, the improvements in CNR and ENL from similar ROIs are com- Raster ORL 1.90 3.76 3.76 9.47 parable with what we achieved in lab tests. Circular NFL 2.85 4.93 6.15 17.03 4 Discussion Circular GCL 1.99 3.91 10.50 22.13 This study implemented, calibrated, and tested a scanning Circular ORL 1.76 3.40 4.52 8.20 modulation technique for speckle reduction in vis-OCT. We addressed unique engineering constraints of clinical vis-OCT, including slower image acquisition speed, intrinsically reduced SNR, and the need to preserve structurally localized, high- 3.4 Speckle Reduction Test in Clinical Environment detailed retinal information. As a comparison in near-infrared A clinical photographer without technical knowledge of the (NIR) OCT, good SNR can be achieved at higher imaging scanning protocol independently verified speckle reduction in speeds, direct B-scan averaging is more feasible, and lower res- the human retina using a commercial vis-OCT system. The olutions are less affected by image blurring. In our human im- photographer acquired images using the same procedure as aging tests, where motions were high, direct B-scan averaging in acquiring normal raster scan images and received no additional vis-OCT proved unreliable for producing high-quality images at Neurophotonics 041107-9 Oct–Dec 2019 Vol. 6(4) Rubinoff et al.: Speckle reduction in visible-light optical coherence. . . Fig. 8 Speckle-reduction test in human retina in clinical environment using modulated raster scan. (a) Reference raster B-scan image; (b) the corresponding srB-scan image; (c) magnified view of the region highlighted in panel (a); and (d) magnified view of the region highlighted in panel (b). The five anatomical layers, IS/OS, COST, ROST, RPE, and BM, are labeled. All images are plotted with identical color bar. an A-line rate of 25 kHz, as shown in Figure 6. Our rectangular The axial component of the scanner velocity induces Doppler modulation of the scanning beam reliably produced speckle- shifts to the interference fringe, thereby reducing the image reduced images in both mouse and human retinas without image SNR. To overcome this challenge, Szkulmowski et al. sug- blurring. gested minimizing camera exposure time (demonstrated at 5 μs) The ultimate target of vis-OCT retinal imaging is the clinical to reduce the effects of fringe washout. Such reduction of cam- adoption, where increased resolution and spectroscopic analysis era exposure time, however, would severely compromise image can improve the management of a variety of retinal diseases. quality in vis-OCT, which typically requires camera exposures Therefore, a primary design constraint of our speckle-reduction of ∼40 μs to achieve sufficiently high SNR in the human retina. technique was the usability by photographers in the clinical Our scanning technique does not introduce additional fringe environment, where advanced vis-OCT engineering skills are washout, as the scanning beam moves in discrete steps and unavailable. It is unreasonable to expect a clinical photographer is stationary during each camera exposure. Szkulmowski et al. to make optical adjustments to the system or troubleshoot tech- experienced ∼6-dB loss to peak signal-to-noise-ratio (PSNR) nical issues that arise. We implemented our speckle reduction in scan-optimized images at an exposure time of 20 μs, which in a clinical vis-OCT system at NYU Langone Medical Center they attribute to fringe washout. We did not observe PSNR drop simply via a software update. No additional optical calibrations in our scan-optimized images at an exposure time of 40 μs. were made to the clinical system, and no changes were made to Although our discrete scanning trajectory is optimized for vis- the photographers’ imaging protocol. Critically, the photogra- OCT, it is expected to provide superior image quality in other phers achieved comparable image quality (Fig. 8) to experts OCT systems with higher exposure times when using the same in a controlled lab environment and the clinical images showed sampling parameters as a resonance mirror-based technique. delineation of RPE and BM [Figs. 8(b) and 8(d)]. Micrometer- Our scanning technique also addresses the trajectory limits scale basal linear and basal laminar deposits between RPE and of a single resonant mirror shown in Szkulmowski et al. Using BM are thought to be early indicators of macular degeneration, discrete, software-controlled XY galvanometer scanners, we a leading cause of blindness. A future study in which these enabled multiple modulation waveforms across multiple coordi- morphological changes are observed in vivo in the clinic may nate systems (e.g., Cartesian and polar). There is promise for open a new window for the diagnosis and management of applying circular scan modulation for circumpapillary retinal macular degeneration. Such capabilities have been previously oximetry, where speckles in blood may disrupt true spectro- unavailable in clinical NIR OCT systems due to reduced axial scopic signal. The velocity of blood (e.g., ∼0.014 μm∕μs for resolution. Presently, NIR OCT is limited to imaging larger scale a 100-μm diameter human retinal artery ) is not high enough drusen, which are developed only at a more advanced stage of to completely uncorrelate the local scattering structure of eryth- macular degeneration. rocytes during consecutive A-lines (period of 40 μs). Therefore, Szkulmowski et al. performed averaging of offset A-lines A-line averaging across a regular circular scan is not perfectly in NIR OCT using a resonant scanner. One hardware challenge efficient for reducing the effects of speckle. By scanning a larger expressed in this study was fringe washout. In addition to the volume, our modulated circular scan is expected to acquire more complication of adding hardware to the system, the resonant uncorrelated speckle patterns in blood using the same number of scanner continuously moves during a single camera exposure. A-lines as a regular circular scan. Furthermore, the modulated Neurophotonics 041107-10 Oct–Dec 2019 Vol. 6(4) Rubinoff et al.: Speckle reduction in visible-light optical coherence. . . scan pattern is not expected to disrupt the SNR of the blood CNR and 2.87-fold increase in ENL can be achieved. We also signal since it does not induce additional fringe washout. showed that reducing speckle via direct B-scan averaging is not Next, by selecting a rectangular waveform, we best preserved as reliable for preserving the fine features, due to retinal motion. the lateral resolution along the B-scan axis and took full advan- Our speckle-reduction method offered a local sampling rate of tage of the space available along the orthogonal axis for speckle 3125 Hz (Sec. 2.1), which is well beyond the motion frequency. pattern decorrelation. In comparison, a resonant scanner can In the future, we will further improve the speckle-reduction only oscillate along a single axis. Furthermore, the reported res- performance, especially in clinics, by providing real-time feed- onant scanner uses a sinusoidal waveform, in which each mirror backs of CNR and ENL so that the photographer can adjust deflection along the orthogonal axis also carries a component imaging parameters to achieve optimal image quality for differ- along the B-scan axis. In this case, averaging consecutive ent eye conditions. In addition to exploring clinical benefits of A-lines will reduce the lateral resolution along the B-scan axis. discriminating minute anatomical features, such as the RPE and In addition, the sinusoidal trajectory makes orthogonal step size, BM, we will also investigate whether circular scan modulation d, nonlinear across each scan of the orthogonal axis. Our find- can improve the accuracy of measuring retinal oxygen saturation ings suggest that image quality is optimized when d is approx- because an srB-scan samples a larger retinal volume than a imately equal to the focused spot diameter along the orthogonal regular B-scan image. axis. Precise optimization is not possible with a nonlinear trajectory. Szkulmowski et al. partially overcame this by only Disclosures acquiring A-lines along the pseudolinear region of every other R.K., Y.W., J.S.S., and H.F.Z. have financial interests in sinusoidal edge, wasting a portion of light exposure on the Opticent Health, which did not support this work. Other authors retina. However, this is still less efficient than scanning with declare no conflicts of interest related to this article. a rectangular wave, in which all spaces along the orthogonal axis are linear and can be fully utilized. Acknowledgments Acquiring A-lines along the orthogonal axis increases imag- This work was supported in part by NIH Grants Nos. ing time by a factor of n ¼ 8. This limits the total number of R01EY026078, DP3DK108248, R01EY029121, R01EY028304, srB-scans per acquisition to 16 for an imaging time of ∼5s. R44EY026466, and T32EY25202. L.B. was supported by the Naturally, 16 raster srB-scans are not feasible to reconstruct a NSF Graduate Research Fellowship 1000260620. The authors high-quality en-face image of the human retina. A potential sol- would like to thank David Miller and Rozita Ghassabi for their ution is to decompose each srB-scan with n ¼ 8 into eight regu- helpful discussions. lar B-scans orthogonal axis, each of 1024 A-lines. If the spacing between each srB-scan is d, then we can decompose 16 srB- scans into 128 equidistant regular B-scans separated by the References distance d. By reducing the number of srA-lines per srB-scan 1. D. Huang et al., “Optical coherence tomography,” Science 254, to 4096 at n ¼ 8, we can decompose the full raster scan into 1178–1181 (1991). 512 A-lines −256 B-scans, and so on. The user may view regu- 2. M. R. Hee et al., “Optical coherence tomography of the human retina,” lar B-scans or srB-scans at a location of interest in the en-face Arch. Ophthalmol. 113, 325–332 (1995). image. Rather than choosing between a high-quality en-face 3. M. Wojtkowski, “High-speed optical coherence tomography: basics and image and high-quality B-scans, the user may adjust scanning applications,” Appl. Opt. 49, D30–D61 (2010). 4. J. M. Schmitt, S. H. Xiang, and K. M. Yung, “Speckle in optical coher- parameters to acquire both simultaneously. We recognize, how- ence tomography,” J. Biomed. Opt. 4(1), 95–105 (1999). ever, that the aforementioned solution is still not as optimal as a 5. M. R. Hee et al., “Optical coherence tomography of age-related macular regularly raster-scanned en-face image due to motion and that degeneration and choroidal neovascularization,” Ophthalmology 103, extra image registrations may be required. 1260–1270 (1996). 6. M. R. Hee et al., “Optical coherence tomography of central serous cho- rioretinopathy,” Am. J. Ophthalmol. 120,65–74 (1995). 5 Conclusion 7. A. Ozcan et al., “Speckle reduction in optical coherence tomography images using digital filtering,” J. Opt. Soc. Am. A 24, 1901–1910 We showed that our orthogonal modulation protocol signifi- (2007). cantly reduced the speckle for both raster and circular scans 8. M. Bashkansky and J. Reintjes, “Statistics and reduction of speckle in in vis-OCT without additional hardware. This protocol was optical coherence tomography,” Opt. Lett. 25, 545–547 (2000). robust against retinal motions. Furthermore, we showed that 9. A. Sakamoto, M. Hangai, and N. Yoshimura, “Spectral-domain optical speckle-reduced vis-OCT imaging does not require additional coherence tomography with multiple B-scan averaging for enhanced training for a clinical photographer to operate, allowing smooth imaging of retinal diseases,” Ophthalmology 115, 1071–1078 (2008). 10. M. Szkulmowski et al., “Efficient reduction of speckle noise in optical clinical translation. We designed this orthogonal modulation coherence tomography,” Opt. Express 20, 1337–1359 (2012). protocol to be compatible with arbitrary scanning trajectories 11. A. E. Desjardins et al., “Speckle reduction in OCT using massively- and demonstrated such capability in circular scans for the first parallel detection and frequency-domain ranging,” Opt. Express 14, time. 4736–4745 (2006). In the mouse retinas, we showed that up to a 2.35-dB (116%) 12. M. Pircher et al., “Speckle reduction in optical coherence tomography improvement in CNR and 3.1-fold increase in ENL can be by frequency compounding,” J. Biomed. Opt. 8(3), 565–569 (2003). 13. Y. Zhao et al., “Real-time speckle reduction in optical coherence tomog- achieved. We recognized that an anesthetized and stabilized raphy using the dual window method,” Biomed. Opt. Express 9, mouse retina had negligible motion, allowing for simple B-scan 616–622 (2018). averaging, and further conducted human retinal imaging. 14. O. Liba et al., “Speckle-modulating optical coherence tomography in Consistently improved image qualities after speckle reduction living mice and humans,” Nat. Commun. 8, 15845 (2017). were shown in human retinas using both an experimental and 15. C. Cuartas-Velez et al., “Volumetric non-local-means based speckle a clinical vis-OCT system and provide comparable results. reduction for optical coherence tomography,” Biomed. Opt. Express Here, we showed that up to a 2.25-dB (94%) improvement in 9, 3354–3372 (2018). Neurophotonics 041107-11 Oct–Dec 2019 Vol. 6(4) Rubinoff et al.: Speckle reduction in visible-light optical coherence. . . University, and a practicing ophthalmologist at Northwestern 16. X. Shu, L. Beckmann, and H. F. Zhang, “Visible-light optical coherence Memorial Hospital. She previously worked at the Jules Stein Eye tomography: a review,” J. Biomed. Opt. 22(12), 121707 (2017). Institute at UCLA and the Doheny Retina Institute at USC. Her 17. J. Yi et al., “Visible-light optical coherence tomography for retinal research interests include translational approaches to age-related oximetry,” Opt. Lett. 38, 1796–1798 (2013). macular degeneration and ischemic retinal diseases. 18. J. Yi et al., “Human retinal imaging using visible-light optical coherence tomography guided by scanning laser ophthalmoscopy,” Biomed. Opt. Xiaorong Liu is an assistant professor of biology and psychology Express 6, 3701–3713 (2015). at the University of Virginia. She received her PhD from the 19. S. P. Chong et al., “Ultrahigh resolution retinal imaging by visible light University of Virginia in 2002 and received postdoctoral training at OCT with longitudinal achromatization,” Biomed. Opt. Express 9, the Scripps Research Institute and the University of California, San 1477–1491 (2018). Francisco. Her research interests include the structural and functional 20. S. Martinez-Conde, S. L. Macknik, and D. H. Hubel, “The role of development of retinal ganglion cells and how they degenerate in fixational eye movements in visual perception,” Nat. Rev. Neurosci. glaucoma. 5, 229–240 (2004). 21. D. C. Adler, T. H. Ko, and J. G. Fujimoto, “Speckle reduction in optical Jenna Tauber is an MD candidate at the New York University (NYU) coherence tomography images by use of a spatially adaptive wavelet School of Medicine. She received her BS degree in communications filter,” Opt. Lett. 29, 2878–2880 (2004). from Cornell University. 22. B. T. Soetikno et al., “Visible-light optical coherence tomography oxi- metry based on circumpapillary scan and graph-search segmentation,” Katie Jones is a clinical photographer and lab supervisor at the Biomed. Opt. Express 9, 3640–3652 (2018). Advanced Ophthalmic Imaging Laboratory, NYU. She graduated in 2013 with a BS degree in biology from the University of Pittsburgh. 23. J. C. Booij et al., “The dynamic nature of Bruch’s membrane,” Prog. Retinal Eye Res. 29,1–18 (2010). Hiroshi Ishikawa is a professor in the Department of Ophthalmology, 24. C. A. Curcio et al., “The oil spill in ageing Bruch membrane,” Br. J. NYU and administrative director of the Ophthalmic Imaging Center. Ophthalmol. 95, 1638–1645 (2011). He received his MD from Mie University in Japan in 1989. 25. M. Guizar-Sicairos, S. T. Thurman, and J. R. Fienup, “Efficient subpixel image registration algorithms,” Opt. Lett. 33, 156–158 (2008). Joel S. Schuman is a professor and chairman of the Department of 26. S. H. Yun et al., “Motion artifacts in optical coherence tomography with Ophthalmology, NYU. He received his MD from Mt. Sinai School of frequency-domain ranging,” Opt. Express 12, 2977–2998 (2004). Medicine in 1984 and served as professor and chairman of the 27. C. E. Riva et al., “Blood velocity and volumetric flow-rate in human Department of Ophthalmology at the University of Pittsburgh from retinal-vessels,” Invest. Ophthalmol. Visual Sci. 26, 1124–1132 (1985). 2003 to 2016. He is a coinventor of OCT and a pioneer in the study and treatment of glaucoma. Ian Rubinoff is a second-year PhD student in biomedical engi- neering at the Functional Optical Imaging Lab (FOIL) at Northwestern Roman Kuranov is the head of product engineering and develop- University. In 2017, he graduated from Lehigh University with a ment at Opticent Health in Evanston, Illinois, and a research specialist BS degree in electrical engineering and physics. His research inter- at Northwestern University. He received his PhD in laser physics ests include innovating visible-light optical coherence tomography from the Institute of Applied Physics of the Russian Academy of (vis-OCT) technology, optics, and signal processing. Sciences and did his postdoctoral training at the University of Texas, Galveston. He previously worked as principal scientist in Wasatch Lisa Beckmann is a third-year PhD student in biomedical engineer- Photonics and as an instructor and researcher at the University of ing at the FOIL at Northwestern University. In 2016, she graduated Texas, San Antonio. from the California Institute of Technology in biomedical engineering. Her research interests include using vis-OCT to study eye patholo- Hao F. Zhang is a professor in the Department of Biomedical gies, optics, and animal model imaging. Engineering at Northwestern University. He leads the FOIL, which seeks to develop and apply new optical imaging technologies for Yuanbo Wang is a senior project engineer at Opticent Health in biological study and clinical use. His lab pioneered the development Evanston, Illinois. He graduated with a PhD in biomedical engineering of vis-OCT for high-resolution and functional retinal imaging. He from the University of Missouri-Columbia, where he studied polariza- received his PhD in biomedical engineering from the Texas A&M tion sensitive OCT and fiber tractography. University in 2006 and received his postdoctoral training from the Department of Biomedical Engineering at Washington University in Amani A. Fawzi is the Cyrus Tang and Lee Jampol Professor of St. Louis. Zhang also cofounded Opticent Health, which seeks to Ophthalmology at the Feinberg School of Medicine, Northwestern translate vis-OCT to the clinic. Neurophotonics 041107-12 Oct–Dec 2019 Vol. 6(4) http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Neurophotonics SPIE

Speckle reduction in visible-light optical coherence tomography using scan modulation

Loading next page...
 
/lp/spie/speckle-reduction-in-visible-light-optical-coherence-tomography-using-GcUcjtAFWW

References (35)

Publisher
SPIE
Copyright
© The Authors. Published by SPIE under a Creative Commons Attribution 4.0 Unported License. Distribution or reproduction of this work in whole or in part requires full attribution of the original publication, including its DOI.
ISSN
2329-423X
eISSN
2329-4248
DOI
10.1117/1.NPh.6.4.041107
Publisher site
See Article on Publisher Site

Abstract

Speckle reduction in visible-light optical coherence tomography using scan modulation Ian Rubinoff Lisa Beckmann Yuanbo Wang Amani A. Fawzi Xiaorong Liu Jenna Tauber Katie Jones Hiroshi Ishikawa Joel S. Schuman Roman Kuranov Hao F. Zhang Ian Rubinoff, Lisa Beckmann, Yuanbo Wang, Amani A. Fawzi, Xiaorong Liu, Jenna Tauber, Katie Jones, Hiroshi Ishikawa, Joel S. Schuman, Roman Kuranov, Hao F. Zhang, “Speckle reduction in visible-light optical coherence tomography using scan modulation,” Neurophoton. 6(4), 041107 (2019), doi: 10.1117/1.NPh.6.4.041107. Neurophotonics 6(4), 041107 (Oct–Dec 2019) Speckle reduction in visible-light optical coherence tomography using scan modulation a a b c d e e Ian Rubinoff, Lisa Beckmann, Yuanbo Wang, Amani A. Fawzi, Xiaorong Liu, Jenna Tauber, Katie Jones, e e a,b a,c, Hiroshi Ishikawa, Joel S. Schuman, Roman Kuranov, and Hao F. Zhang * Northwestern University, Department of Biomedical Engineering, Evanston, Illinois, United States Opticent Health, Evanston, Illinois, United States Northwestern University, Department of Ophthalmology, Chicago, Illinois, United States University of Virginia, Department of Biology and Psychology, Charlottesville, Virginia, United States New York University, Department of Ophthalmology, New York, United States Abstract. We present a technique to reduce speckle in visible-light optical coherence tomography (vis-OCT) that preserves fine structural details and is robust against sample motion. Specifically, we locally modulate B-scans orthogonally to their axis of acquisition. Such modulation enables acquisition of uncorrelated speckle patterns from similar anatomical locations, which can be averaged to reduce speckle. To verify the effectiveness of speckle reduction, we performed in-vivo retinal imaging using modulated raster and circular scans in both mice and humans. We compared speckle-reduced vis-OCT images with the images acquired with unmodulated B-scans from the same anatomical locations. We compared contrast-to-noise ratio (CNR) and equivalent num- ber of looks (ENL) to quantify the image quality enhancement. Speckle-reduced images showed up to a 2.35-dB improvement in CNR and up to a 3.1-fold improvement in ENL with more discernable anatomical features using eight modulated A-line averages at a 25-kHz A-line rate. © The Authors. Published by SPIE under a Creative Commons Attribution 4.0 Unported License. Distribution or reproduction of this work in whole or in part requires full attribution of the original publication, including its DOI. [DOI: 10.1117/1.NPh.6.4.041107] Keywords: optical coherence tomography; visible light; speckle; retina; imaging; clinic. Paper 19045SSR received May 13, 2019; accepted for publication Aug. 8, 2019; published online Sep. 3, 2019. uncorrelated speckle patterns. Uncorrelated patterns from sim- 1 Introduction ilar structural locations can be averaged to remove the speckle Optical coherence tomography (OCT) is a scattering-based and reveal the original anatomical information. The physical imaging technology that acquires high-resolution three- 1 basis of incoherent averaging method makes it ideal for situa- dimensional images of biological samples in vivo. Following tions where the study of fine anatomical features is required. its initial report in 1991, OCT has become the “gold standard” However, manipulating image acquisition to obtain uncorrelated for noninvasive retinal imaging. Today, it is an essential technol- speckle patterns can be challenging. First, different scattering ogy in labs and clinics for studying and managing a wide variety 2 events must be probed without losing the structural integrity of retinal diseases. Advances in optoelectronics in the past of the location of interest. Second, multiple acquisitions at a par- 25 years has led to improved resolution, signal-to-noise ratio ticular location are required to generate enough patterns suitable (SNR), and imaging field of view (FOV) in OCT. However, for averaging. Samples with strong motion can pose a challenge speckle, an image artifact caused by the self-interference of to averaging, especially in human eyes. Multiple approaches coherent light at random phases, remains a significant source 4 have been developed to achieve incoherent averaging of speckle of reduced image quality. This is of particular salience in retinal while retaining high image quality. The most basic technique is imaging, where speckle noise can obscure fine structures in the to average consecutive B-scans in a raster pattern, either from outer retina, such as the retinal pigment epithelium (RPE) and the same location or from a slightly offset position. The former Bruch’s membrane (BM). Minute pathological changes in these relies on a small sample movement to modify scattering events, structures may be strongly associated with the progressions of and the latter directly modifies scattering events across consecu- several retinal diseases, including macular degeneration and tive separated B-scans. More advanced techniques for spatial central serous retinopathy (CSR). averaging include modulating the scanning beam after every To improve the imaging quality, researchers have developed 10 11 A-line with a translational offset or angular offset. In particu- several methods to suppress speckle artifacts in OCT. These lar, translational offset has been shown to be more robust against methods can be classified into two categories: digital filtering sample motion than B-scan averaging. Both scan modulations and incoherent averaging. Digital filtering, while simple to have been previously implemented with additional hardware implement and effective in reducing the grainy appearance of and moving parts. Other techniques include multiwavelength speckle, causes blurring that degrades image resolution and pre- 12,13 14 averaging, modulation of light wavefront, and nonlocal, vents the delineation of fine anatomical features. Incoherent software-based averaging. averaging, on the other hand, samples photons that have under- Recent development of visible-light optical coherence gone statistically different scattering events, thereby generating tomography (vis-OCT) has generated new capabilities for retinal imaging, including visualization of fine structures with ultrahigh resolution and spectroscopic analysis of blood-oxygen concen- *Address all correspondence to Hao F. Zhang, E-mail: hfzhang@north 16–18 western.edu tration (sO ). Speckle, which distorts both structural and Neurophotonics 041107-1 Oct–Dec 2019 Vol. 6(4) Rubinoff et al.: Speckle reduction in visible-light optical coherence. . . spectral information, can undermine these new benefits. To reduce speckle in vis-OCT using incoherent averaging, there are some unique challenges. First, there is strong optical absorption and scattering in tissue in the visible-light spectral range, reduc- ing the amount of photons that can be collected per unit time. This is coupled with high relative intensity from the super- continuum laser source. To achieve high SNR, a prolonged camera exposure is required, therefore reducing scanning speed. Second, since eyes are sensitive to visible-light illumination, vis-OCT often suffers from increased retinal motion. Finally, preservation of structure-dependent spectroscopic information requires anatomically localized sampling of scattering events. A method for speckle reduction that is both structurally local- ized and robust against retinal motion is optimal for vis-OCT. Furthermore, it is ideal for a speckle reduction technique to be easily implementable in clinical settings to increase usability, reduce likelihood of malfunction, and cause no additional discomfort to patients. A straightforward approach is to perform B-scan averaging. However, eye motion in vis-OCT makes B-scan averaging unreliable due to blurring, even with post- acquisition image registration. To overcome this challenge, we modulated the scanning trajectory orthogonally to the direction of the B-scan images during data acquisition. This method Fig. 1 Illustrations of speckle-reduction scanning protocols. enabled sampling of different speckle patterns while maintain- (a) Overall illustration of the relationship between the B-scan axis and the orthogonal axis in the modulated raster scan; (b) detailed illustra- ing high anatomical similarity between modulations. We imple- tion of the A-line acquisition sequence in the modulated raster scan. mented scanning modulation by directly controlling the Here, d is the distance between two adjacent A-lines along the galvanometer scanners without additional hardware. orthogonal axis. The arrows 1, 2, and 3 highlight the trajectory of galvanometer motion. (c) Overall illustration of the relationship be- tween the B-scan axis and the orthogonal axis in the modulated cir- 2 Methods cular scan; (d) detailed illustration of the A-line acquisition sequence in the modulated circular scan. 2.1 Scanning Protocol We modulated both raster and circular scans, which are com- We averaged all n A-lines in the orthogonal direction along monly used in vis-OCT, to test our speckle reduction method. each rectangular edge [Figs. 1(b) and 1(d)] to generate a single Figure 1 illustrates the modulated raster scan [Figs. 1(a) and speckle-reduced A-line (srA-line). For a desired sampling den- 1(b)] and modulated circular scan [Figs. 1(c) and 1(d)]. As sity of m srA-lines per speckle-reduced B-scan (srB-scan), the shown in Figs. 1(a) and 1(c), we define the B-scan axis as the total number of camera acquisitions per srB-scan is n × m. Each direction along which a traditional cross-sectional image would consecutive srA-line in an srB-scan can then be calculated as be acquired without modulation. We define the orthogonal axis as the direction orthogonal to the B-scan axis on the two- dimensional scanning plane. Movement along the orthogonal EQ-TARGET;temp:intralink-;e001;326;328srA ¼ A for j ¼ 0; 1; 2;:::m − 1; (1) j ij axis [arrow 1 in Figs. 1(b) and 1(d)] occurs in n equidistant i¼1 steps, where n is the number of speckle-uncorrelated A-lines where i is the index of each set of n A-lines about the edge of a to be locally averaged. Each translation of the galvanometer [red dots in Figs. 1(b) and 1(d)] is discrete, synchronized with the rectangle; j is the index of each consecutive edge of a rectangle; spectrometer camera exposure, and implemented entirely via and A is the i þ j × n’th A-line in a full B-scan acquisition. ij software control without additional hardware or moving parts. Since an srB-scan increases imaging time over normal This avoids a complex synchronization procedure or risk of B-scan acquisition by a factor of n, it is important to collectively desynchronization between the beam path and the camera expo- limit n, m, and the number of total srB-scans to prevent overly sure when using an external scanner. The centroids of each long imaging time. First, all sampling numbers were selected spot generating an A-line are separated by a distance d along in powers of 2 to support fast graphics processing unit data the orthogonal axis [Fig. 1(b)]. After n translations in this direc- processing. Next, we limited all imaging experiments to 8192 tion, the beam is shifted along the B-scan axis [arrow 2 in total A-lines per srB-scan. Given a camera exposure time of Figs. 1(b) and 1(d)], followed by a reversed scan along the 40 μs, which is required for sufficiently high SNR, an srB-scan orthogonal axis [arrow 3 in Figs. 1(b) and 1(d)]. Such modula- could be acquired in 328 ms, an upper limit for reducing bulk tion superimposes a rectangular wave on the B-scan axis, where motion artifacts (satisfying Nyquist criterion of 500 ms for eye each rising and falling edge of each rectangle contains n microsaccades of ∼1Hz). Furthermore, we chose to limit the speckle-uncorrelated A-lines. While other modulation shapes total image acquisition time to ∼5s to prevent patient fatigue such as sinusoidal or triangular are possible, we chose rectan- and discomfort. This limited the total amount of srB-scans per gular to best preserve lateral resolution along the B-scan axis. acquisition to 16 (5.25 s total acquisition time). In our experi- During acquisition, several parameters, including n, d, and mental human imaging system (Sec. 2.3.2), we maximized imaging FOV are adjustable. We investigate how to obtain an lateral sampling density in a raster scan without spot overlap, optimal d in Sec. 3.1. where m ¼ 1024 srA-lines and n ¼ 8 averages. The parameters Neurophotonics 041107-2 Oct–Dec 2019 Vol. 6(4) Rubinoff et al.: Speckle reduction in visible-light optical coherence. . . n, m, and d can be easily modified for different experimental between the reference B-scan and srB-scans to evaluate the conditions. effectiveness of speckle reduction. The srB-scan averages along the same locations as n spatially separated B-scans, each of m A-lines. However, in direct B-scan 2.3 Data Acquisition averaging, each A-line at a particular lateral position is delayed temporally by the scanner’s fly-back time. This results in a total We tested our speckle reduction protocol in both mouse and sampling period of n × m × t, where t is the camera exposure human retinas using two prototype systems developed at time for each A-line. In our method, modulation removes the Northwestern University. In addition, we further tested our wait for scanner fly-back, thereby reducing the total sampling speckle reduction method in humans in a clinical setting using period to n × t for each srA-line. Using our experimental param- a commercial vis-OCT system (Aurora X1, Opticent Health), eters (m ¼ 1024, n ¼ 8, t ¼ 40 μs), we reduced the sampling where optical engineering expertise was unavailable. We period from 328 ms to 320 μs for each srA-line and increased directly implemented the modulated scanning protocol in that the srA-line rate from 3 to 3125 Hz. Constant, involuntary reti- system without additional calibration, alignment, or changes nal motions can occur at frequencies up to 90 Hz with ampli- to the photographer’s workflow. tudes up to 40 arc sec (equivalent to 0.011-μm change in sampling location per 40-μs camera exposure in the human 2.3.1 Mouse imaging retina). This leaves the possibility of only 0.088 μm of move- ment during an srA-line, which is insignificant when compared For mouse imaging, we used the system described in our pre- 22 2 with the micron-order lateral and axial resolutions in OCT. vious work. In brief, a 1∕e spot size of ∼5.7 μm was incident Therefore, the improved srA-line rate is highly significant. on the retina. We controlled the total illumination power to 1.2 mW on the cornea in all instances. For a raster scan, we used m ¼ 1024, n ¼ 8, and d ¼ 6.3 μm.AnFOVof 1.4 × 1.4 mm 2.2 Metrics to Evaluate Image Quality Improvement was used in all mouse retina images, equivalent to ∼1.4-μm We used contrast-to-noise ratio (CNR) and equivalent number separation between srA-lines along the B-scan axis. For a cir- of looks (ENL) to evaluate image quality improvement after cular scan, we used n ¼ 8 and d ¼ 5.4 μm. The circle circum- speckle reduction. CNR measures how well the sample feature ference was 1.8 mm, equivalent to ∼1.8-μm separation between can be discerned from the surrounding background. Mean inten- srA-lines along the B-scan axis. Improved from our previous sity and variance from both the image background and the system, we adopted a commercial spectrometer (Blizzard SR, sample feature are included to account for two separate noise Opticent Health) with a 2048-pixel line scan camera covering components: intrinsic OCT background noise and speckle. 510 to 610 nm, which provided an axial resolution of ∼1 μm Since the optical properties of different features vary, we calcu- in tissue. We used an A-line rate of 25 kHz in all rodent lated the CNR (dB) in confined region of interests (ROIs) as experiments. All rodent experimental procedures were approved by the μ − μ i b Northwestern University IACUC and conformed to the Asso- EQ-TARGET;temp:intralink-;e002;63;403CNR ¼ 10 log pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ; (2) 2 2 σ þ σ ciation for Research in Vision and Ophthalmology (ARVO) i b Statement on Animal Research. We anesthetized adult where μ is the mean intensity of the i’th ROI; μ is the mean i b C57Bl6/J mice (n ¼ 8) with an intraperitoneal injection intensity of the background outside of the sample structure; σ is (10 mL∕kg body weight) of a ketamine/xylazine cocktail the variance of the i’th ROI; and σ is the variance of the back- (ketamine: 11.45 mg∕mL; xylazine: 1.7 mg∕mL). Each mouse ground outside of the sample feature. was then placed on a custom-made animal holder and immobi- ENL is the squared inverse of the speckle contrast and mea- lized for imaging. The body temperature was maintained with sures the smoothness and homogeneity within an ROI. We cal- a heat lamp. To dilate the pupil, we applied a drop of 1% culated ENL as tropicamide hydrochloride ophthalmic solution. Throughout imaging, we applied one drop of commercial artificial tears after each image acquisition to prevent corneal dehydration. EQ-TARGET;temp:intralink-;e003;63;283ENL ¼ ; (3) After the imaging session concluded, the mouse was allowed to recover under heat lamp and was returned to the animal where μ is the mean intensity of the i’th ROI and σ is the vari- housing facility. ance of the i’th ROI. An increase in ENL serves as a strong indicator for the reduction of speckle. 2.3.2 Human imaging We compared CNR and ENL in srB-scans with a “reference” B-scan from the same location as an srB-scan. A reference Human imaging was performed using two vis-OCT systems. B-scan included 8192 A-lines acquired along the B-scan axis First, images were acquired in the Ophthalmology Department with a 40-μs camera exposure. Every eight consecutive A-lines at Northwestern Memorial Hospital using an experimental sys- were averaged, resulting in a final sampling density of 1024 tem reported in our previous work. We controlled the illumi- averaged A-lines per reference B-scan. This operation was nation power to be <250 μW on the cornea in all our human equivalent to acquiring an srB-scan without modulating the studies. A 1∕e spot size of ∼6.3 μm was incident on the retina. scanner along the orthogonal axis. Because of high sampling For a raster scan, we used n ¼ 8, d ¼ 7 μm, and m ¼ 1024. The density along the B-scan axis, averaged speckle patterns were FOV was 6.8 × 6.8 mm , leading to ∼6.6-μm separation be- still highly correlated, preventing reduction of speckle. tween srA-lines along the B-scan axis. For a circular scan, However, background noise was equally suppressed in reference we used n ¼ 8 and d ¼ 5.9 μm. The circle circumference was B-scans and srB-scans because they used the same amount of 18.3 mm, equivalent to ∼18-μm separation between srA-lines temporal averaging. We compared the CNR and ENL values along the B-scan axis. Similar to our mouse system, we Neurophotonics 041107-3 Oct–Dec 2019 Vol. 6(4) Rubinoff et al.: Speckle reduction in visible-light optical coherence. . . upgraded to a commercial spectrometer (Blizzard SR, Opticent Figures 2(g)–2(j) show the similar comparison in the circular Health). The A-line rate was 25 kHz in all human imaging scan, where the speckle-reduced circular scan with d ¼ 6 μm tests. demonstrates improvement in image quality. To verify the efficacy of our method outside the lab environ- Figures 2(k) and 2(l), respectively, show the pixel intensity ment, we further conducted human imaging in the Department histograms from the tape layer 1 [highlighted in Fig. 2(c)] in the of Ophthalmology at New York University (NYU) Langone raster and circular scans. In both scan patterns, the intensity his- Medical Center. A clinical vis-OCT system (Aurora X1, tograms changed from a broad, right-skewed distribution when Opticent Health) was used to acquire all images. It offered d ¼ 0 μm to a lower-variance, nearly centrosymmetric distribu- an axial resolution of ∼1 μm and we controlled the spectrometer tion when d ¼ 6 μm. These results agree with the expected exposure time to be 40 μs. We used the same raster scanning change in pixel intensity distribution from Rayleigh distribution parameters as those in the Northwestern system but reduced the to Poisson distribution after speckle reduction. FOV to 5 × 5mm . We implemented the speckle-reduction As shown in Figs. 2(a) and 2(b), we used a −0.25-dB drop acquisitions in Aurora X1 entirely via a software update without in CNR to determine the range of acceptable d values, which any additional calibration or hardware modifications. Clinical gives d ¼ 5.1 μm and d ¼ 7.8 μm. This range is helpful min max photographers acquired retinal images without any changes to for human retinal imaging, where the eye shape, optical proper- their normal workflow. ties, and scanning location may differ among subjects. We also All human imaging procedures in the respective imaging noted that the optimal d ¼ 6 μm is approximately equal to the locations were approved by the Northwestern University estimated spot size of ∼5.5 μm on the retina. This suggests Institutional Review Board (IRB) and NYU IRB and adhered that adjacent spots along the orthogonal axis should be as close to the tenets of the Declaration of Helsinki. Healthy volunteers as possible without spatial overlap. This result is consistent with without known eye diseases provided informed consent before the notion that spatial overlapping provides correlated speckle imaging (Northwestern site: n ¼ 3; NYU site: n ¼ 6). patterns. This result also suggests that it is acceptable to estimate the optimal d using the OCT focal spot size on the sample. These considerations are not expected to change in the living 2.4 Initial Calibration for Orthogonal Spot human eye, where local movement during a single srA-line Separation (0.088 μm) is significantly less than d. We adjusted the d value within the identified range in rodent A calibration procedure was needed for coarse determination and human imaging to accommodate different eye conditions. of optimal spot separation, d, along the orthogonal axis. Since Since we control the d value by the galvanometer angle, we CNR is associated with the ability to discern features from noise, identified optimal angular step size along the orthogonal axis we used it as the primary indicator for image quality. In theory, in different experimental conditions. For mouse imaging, the an increased d increases the decorrelation of the speckle patterns optimal angular step sizes were 0.175 deg and 0.15 deg, which between adjacent orthogonal A-lines. After averaging, speckle correspond to d values of 6.3 and 5.4 μm, respectively, in raster is maximally reduced when the averaged patterns are entirely and circular scans. For human imaging, the optimal angular step uncorrelated. However, if d is too large, we will lose structural sizes were 0.025 deg and 0.02 deg, which correspond to d values similarity between orthogonal A-lines, which can result in of 7 and 5.9 μm, respectively, in raster and circular scans. image blurring. We investigated the impact of modulation dis- tance on CNR by imaging a model mouse eye using both raster and circular scans. The model eye was made from a silica bead (diameter: 3.15 mm). We attached two layers of tape and paper 3.2 Speckle Reduction in the Mouse Retina with an ink pattern to the bottom of the bead to simulate the Figure 3 shows the speckle reduction results in a mouse retina retinal layers. Using the rodent vis-OCT system, we reached using a raster scan. Figures 3(a) and 3(b) are the reference a 1∕e spot size of ∼5.5 μm on the tape layers through the bead. B-scan and srB-scan images, respectively. The imaged retina We then varied the d value from 0 to 13.75 μm in 16 steps and in the srB-scan has a smoother, less grainy appearance that pro- acquired an srB-scan after each step. We calculated CNR from vides a clearer differentiation between anatomical layers. We three ROIs in the top tape layer and averaged them to determine selected six ROIs from the inner plexiform layer [IPL, high- the impact of the d value on image quality. lighted by (a1) and (b1)], outer nuclear layer [ONL, highlighted by (a2) and (b2)], and outer retinal layer [ORL, highlighted by 3 Results (a3) and (b3)] to quantify quality improvement. Figures 3(a1)– 3(a3) and Figs. 3(b1)–3(b3) show the magnified views of the six 3.1 Impact of Modulation Distance on Image Quality selected ROIs and Table 1 shows the quantitative comparisons The results to identify an optimal d value are shown in Fig. 2. of CNR and ENL values from these ROIs. Speckle reduction is Figures 2(a) and 2(b), respectively, show how CNR values vary particularly helpful in the ORL, where a small gap near RPE and as a function of d in imaging the model eye using modulated BM layers is revealed [Fig. 3(b3)], which is not visible in the raster and circular scans. When d is increased from 0 to reference B-scan image [Fig. 3(a3)]. The capability to differen- 13.75 μm in both scans, CNR reaches its maximum at d ¼ tiate RPE and BM may add significant value to various preclini- 6 μm. Figures 2(c) and 2(e) show raster srB-scans with d ¼ 0 cal studies using mouse models. and 6 μm, respectively. Figures 2(d) and 2(f) show magnified Figure 4 shows the speckle reduction results in a mouse views of the two highlighted images [yellow boxes in Figs. 2(c) retina using circular scan. Figures 4(a) and 4(b) are the reference and 2(e)], respectively. The srB-scan with d ¼ 6 μm [Fig. 2(f)] B-scan and srB-scan images, respectively. Again, the srB-scan shows a smoother intensity distribution within each layer and image improved the overall image quality with better differen- much improved discrimination between the tape and the paper tiated fine anatomical features. We also selected six ROIs from layers, as compared with the unmodulated scan [Fig. 2(d)]. IPL [highlighted by (a1) and (b1)] and retinal blood vessels Neurophotonics 041107-4 Oct–Dec 2019 Vol. 6(4) Rubinoff et al.: Speckle reduction in visible-light optical coherence. . . Fig. 2 Speckle-reduction test in the model mouse eye. (a) Change of averaged CNR as a function of d in the modulated raster scan; (b) change of averaged CNR as a function of d in the modulated circular scan. (c) An srB-scan image of the model mouse eye acquired using modulated raster scan with d ¼ 0 μm. The structures corresponding to the two tape and one paper layers are highlighted by the arrows. (d) Magnified view of the region highlighted by the box in panel (c). (e) An srB-scan image of the model mouse eye acquired using modulated raster scan with d ¼ 6 μm. (f) Magnified view of the region high- lighted by the box in panel (e). (g) An srB-scan image of the model mouse eye acquired using modulated circular scan with d ¼ 0 μm. (h) Magnified view of the region highlighted by the box in panel (g). (i) An srB-scan image of the model mouse eye acquired using modulated circular scan with d ¼ 6μm. (j) Magnified view of the region highlighted by the box in panel (i). All images are plotted with identical color bar. (k) Fitted pixel-intensity histograms within the tape layer 1 acquired by modulated raster scans with d ¼ 0 and 6 μm; (l) fitted pixel-intensity histograms within the tape layer 1 acquired by modulated circular scans with d ¼ 0 and 6 μm. [highlighted by (a2), (b2), (a3), and (b3), respectively] for quan- For circular scans, the ROIs in the IPL and two vessels show titative evaluation. 1.84, 1.90, and 1.11 dB respective improvement in CNR, and Table 1 compares the CNR and ENL values from the selected 2.69, 2.56, and 1.63 times respective improvements in ENL. ROIs in both raster and circular scans. In each scan mode, we CNR and ENL improvements for the ORL in the raster scan and see increased metric values from the ROIs in the srB-scan second vessel in the circular scan are slightly lower than other images. For raster scans, the ROIs in the IPL, ONL, and ORL improvements. This is because some of the image background is show 2.35, 1.84, and 1.32 dB respective improvements in CNR, unavoidably included in the ROI, artificially contributing low and 3.1, 2.53, and 1.84 times respective improvements in ENL. pixel intensities to μ in the metrics. Neurophotonics 041107-5 Oct–Dec 2019 Vol. 6(4) Rubinoff et al.: Speckle reduction in visible-light optical coherence. . . Fig. 3 Speckle-reduction test in mouse retina using modulated raster scan. (a) Reference raster B-scan image. Three ROIs from IPL, ONL, and ORL are highlighted by (a1), (a2), and (a3), respectively. The size of each ROI is 140 μmðlateralÞ × 20 μmðaxialÞ. (b) The corresponding srB-scan image. The same three ROIs are highlighted by (b1), (b2), and (b3), respectively. (a1)–(a3) The magnified views of the three highlighted ROIs in panel (a). (b1)–(b3) The magnified views of the three highlighted ROIs in panel (b). CNR and ENL values are calculated from all the selected ROIs. All images are plotted with identical color bar. 3.3 Speckle Reduction in the Human Retina RPE, and BM become clearly discernable from one another. The thickness of BM is measured as ∼3 μm and is resolved in the We accomplished speckle reduction in the human retina using whole image without blur or distortion. The average measured both the laboratorial prototype and a clinical vis-OCT system. 23 thickness of BM in the human eye is ∼2 to 5 μm, which is Unlike mouse imaging, in which retinal motion can be mini- consistent with our measurement. The distinct separation mized and images can be acquired over an extended period, between the BM and the RPE, as shown in Fig. 5(a3), may open human imaging usually suffers from severe retinal motions and up new window to investigate macular degeneration, where image acquisition needs to complete within few seconds. For vis- initial pathological alterations are hypothesized to start from 23,24 OCT, retinal motion can be much stronger as described in Sec. 1. BM. Finally, we note a shadow caused by a small blood ves- Figure 5 shows the speckle reduction results using raster scan sel as highlighted by the arrows in both the reference B-scan in a human retina (22-year-old male volunteer). Figures 5(a) and [Fig. 5(a3)] and the srB-scan [Fig. 5(b3)] images in ORL. It 5(b) are reference B-scan and srB-scan images superior to the is measured as 2 pixels laterally or ∼14 μm in width. This fea- optic disk, respectively. The srB-scan is smoother and less ture is better resolved in the srB-scan image, indicating that lat- grainy in appearance than the reference B-scan, increasing vis- eral resolution has been well preserved after speckle reduction. ibility of the retinal layers. Improved image quality here is con- Repetitive B-scan averaging is not trivial due to retinal sistent with that in the mouse retina [Fig. 3(b)]. We selected six motion, which often leads to image blurring even after registra- ROIs from the nerve fiber layer [NFL, highlighted by (a1) and tion. We overcame this challenge and showed that our speckle (b1)], ganglion cell layer [GCL, highlighted by (a2) and (b2)], reduction method is robust against retinal motion in Fig. 6. and ORLs [highlighted by (a3) and (b3)] to quantify quality We acquired eight repeated raster B-scans (each containing improvement. Figures 5(a1)–5(a3) and 5(b1)–5(b3) show the 1024 A-lines) from the same anatomical location and volunteer, magnified views of the six selected ROIs and Table 2 shows the as shown in Figs. 5(a) and 5(b). All the B-scans were axially quantitative comparisons of CNR and ENL values from these and laterally registered using an fast Fourier transform based ROIs. Of particular note is the increased clarity of ORL in the cross-correlation algorithm. The averaged B-scan image srB-scan [Fig. 5(b3)]. Unlike the reference B-scan [Fig. 5(a3)], [Fig. 6(a)] shows blurred anatomical layers in both the inner the shape and boundaries of the rod outer segment tips (ROST), retina and the outer retina due to motion. Figure 6(b) shows Neurophotonics 041107-6 Oct–Dec 2019 Vol. 6(4) Rubinoff et al.: Speckle reduction in visible-light optical coherence. . . Fig. 4 Speckle-reduction test in mouse retina using modulated circular scan. (a) Reference circular B- scan image. Three ROIs from IPL and two vessels are highlighted by (a1), (a2), and (a3), respectively. The size of each ROI is 140 μmðlateralÞ × 20 μmðaxialÞ. (b) The corresponding srB-scan image. The same three ROIs are highlighted by (b1), (b2), and (b3), respectively. CNR and ENL values are calcu- lated from all the selected ROIs. All images are plotted with identical color bar. Table 1 Image quality metric values from the ROIs in the mouse averaged B-scan images acquired by vis-OCT using similar scan retina shown in Figs. 3 and 4. parameters. Figure 6(d) shows a magnified view of the same ana- tomical position from an srB-scan image, where all anatomical layers are clearly resolved across the whole image. The same CNR (dB) CNR (dB) ENL ENL A-line locations from Fig. 6(b) are highlighted in Fig. 6(d) (by Scan type ROI ref. B-scan srB-scan ref. B-scan srB-scan 3 and 4). Figure 6(e) shows A-line 3 and A-line 4, confirming Raster IPL 2.03 4.38 3.94 12.21 that all ORLs are well resolved despite retinal motion. We also demonstrate speckle reduction in circular scan in the Raster ONL 1.63 3.47 3.38 8.56 human retina (Fig. 7). Figures 7(a) and 7(b) show a reference Raster ORL 1.83 3.15 2.76 5.07 B-scan image and an srB-scan image, acquired at the same anatomical location, respectively. We selected six ROIs from Circular IPL 2.32 4.16 4.74 12.74 the same locations as in Fig. 5, including the NFL [highlighted by (a1) and (b1)], GCL [highlighted by (a2) and (b2)], and Circular Vessel 1 1.63 3.53 3.36 8.60 ORL [highlighted by (a3) and (b3)] to quantify quality improve- Circular Vessel 2 2.00 3.11 6.11 9.97 ment. Figures 7(a1)–7(a3) and 7(b1)–7(b3) show magnified views of the six selected ROIs and Table 2 shows the quantita- tive comparisons of the CNR and ENL values from these ROIs. Similar to the ORL in the raster srB-scan [Fig. 6(b3)], the ORL a magnified view of the region highlighted by the box in Fig. 6(a). in the circular srB-scan [Fig. 7(b3)] shows distinct separation Two A-lines from the locations highlighted by lines 1 and 2 in between BM, RPE, and ROST. In the reference B-scan image Fig. 6(b) are shown in Fig. 6(c). A-line 1 reveals five anatomical [Fig. 7(a3)], however, boundaries of these anatomical layers layers in the outer retina, notably with reduced contrast near the are not easily differentiated due to speckles. To the best of our RPE. A-line 2 fails to resolve any anatomical features. Since vis- knowledge, this is the first demonstration of speckle-reduced OCT offers an axial resolution of near 1 μm, small misalignments in B-scan averaging may lead to much severer blurring. The imaging in a circular pattern using localized scan modulation image quality shown in Fig. 6(a) is representative of most in the human retina. Neurophotonics 041107-7 Oct–Dec 2019 Vol. 6(4) Rubinoff et al.: Speckle reduction in visible-light optical coherence. . . Fig. 5 Speckle-reduction test in human retina using modulated raster scan. (a) Reference raster B-scan image. Three ROIs from NFL, GCL, and ORL are highlighted by (a1), (a2), and (a3), respectively. The size of each ROI is 430 μmðlateralÞ × 23 μmðaxialÞ. (b) The corresponding srB-scan image. The same three ROIs are highlighted by (b1), (b2), and (b3), respectively. (a1)–(a3) The magnified views of the three highlighted ROIs in panel (a). (b1)–(b3) The magnified views of the three highlighted ROIs in panel (b). CNR and ENL values are calculated from all the selected ROIs. The bottom three anatomical layers ROST, RPE, and BM are highlighted in panel (b3). The arrows in (a3) and (b3) highlight the same blood vessel shadow. All images are plotted with identical color bar. Fig. 6 Directly comparing averaged B-scan with srB-scan images from human retina. (a) Image scan from the same location as shown in Fig. 5(b) after averaging eight B-scans. (b) Magnified view of the outer retina region as highlighted in panel (a). (c) Two A-lines from the positions highlighted by 1 and 2 in panel (b). (d) Magnified view of the same outer retina region from the srB-scan shown in Fig. 5(b). Five anatomical layers are labeled. (e) Two A-lines from the positions highlighted by 3 and 4 in panel (d). All plotted A-lines are averaged three times laterally to reduce variation. All images are plotted on the same contrast scale as used in Fig. 5. Table 2 compares the CNR and ENL values from the in CNR of 2.08, 1.92, and 1.64 dB, in the NFL, GCL, and selected ROIs in the human retina for both raster and circular ORL, respectively. Here, corresponding ENL improvements scans. In each instance, we see increased metric values in the are 2.77, 2.11, and 1.81 times, respectively. Again, we attrib- srB-scan ROIs. Raster scans show an improvement in CNR of ute the slightly lower metric value increases in the ORL to 2.25,2.00, and1.86dB, in theNFL,GCL,and ORL, respec- the unavoidable inclusion of image background in the ROI, tively. Corresponding ENL improvements are 2.87, 2.72, and artificially contributing low pixel intensities to μ in the 2.52 times, respectively. Circular scans show an improvement metrics. Neurophotonics 041107-8 Oct–Dec 2019 Vol. 6(4) Rubinoff et al.: Speckle reduction in visible-light optical coherence. . . Fig. 7 Speckle-reduction test in human retina using modulated circular scan. (a) Reference circular B- scan image. Three ROIs from NFL, GCL, and ORL are highlighted by (a1), (a2), and (a3), respectively. The size of each ROI is 900 μmðlateralÞ × 40 μmðaxialÞ. (b) The corresponding srB-scan image. The same three ROIs are highlighted by (b1), (b2), and (b3), respectively. (a1)–(a3) The magnified views of the three highlighted ROIs in panel (a). (b1)–(b3) The magnified views of the three highlighted ROIs in panel (b). CNR and ENL values are calculated from all the selected ROIs. The bottom four ana- tomical layers cone outer segment tips (COST), ROST, RPE, and BM are highlighted in panel (b3). All images are plotted with identical color bar. training. Quality improvement in the clinical speckle-reduced Table 2 Image quality metric values from the ROIs in the human images is comparable to the lab tests (Sec. 3.3). Figures 8(a) retina from Figs. 5 and 7. and 8(b), respectively, show a reference B-scan and an srB-scan of the macula from a 37-year-old female volunteer. In the refer- CNR (dB) CNR (db) ENL ENL ence B-scan image, speckle particularly distorts the ORL, as Scan type ROI ref. B-scan srB-scan ref. B-scan srB-scan shown in Fig. 8(c), preventing the delineation of fine anatomical structures, such as the RPE and BM. As a comparison, all five Raster NFL 2.39 4.64 4.19 12.01 ORLs, including the RPE and BM, are clearly resolved in the Raster GCL 2.31 4.31 5.92 16.12 magnified srB-scan image, as shown in Fig. 8(d). Quantitatively, the improvements in CNR and ENL from similar ROIs are com- Raster ORL 1.90 3.76 3.76 9.47 parable with what we achieved in lab tests. Circular NFL 2.85 4.93 6.15 17.03 4 Discussion Circular GCL 1.99 3.91 10.50 22.13 This study implemented, calibrated, and tested a scanning Circular ORL 1.76 3.40 4.52 8.20 modulation technique for speckle reduction in vis-OCT. We addressed unique engineering constraints of clinical vis-OCT, including slower image acquisition speed, intrinsically reduced SNR, and the need to preserve structurally localized, high- 3.4 Speckle Reduction Test in Clinical Environment detailed retinal information. As a comparison in near-infrared A clinical photographer without technical knowledge of the (NIR) OCT, good SNR can be achieved at higher imaging scanning protocol independently verified speckle reduction in speeds, direct B-scan averaging is more feasible, and lower res- the human retina using a commercial vis-OCT system. The olutions are less affected by image blurring. In our human im- photographer acquired images using the same procedure as aging tests, where motions were high, direct B-scan averaging in acquiring normal raster scan images and received no additional vis-OCT proved unreliable for producing high-quality images at Neurophotonics 041107-9 Oct–Dec 2019 Vol. 6(4) Rubinoff et al.: Speckle reduction in visible-light optical coherence. . . Fig. 8 Speckle-reduction test in human retina in clinical environment using modulated raster scan. (a) Reference raster B-scan image; (b) the corresponding srB-scan image; (c) magnified view of the region highlighted in panel (a); and (d) magnified view of the region highlighted in panel (b). The five anatomical layers, IS/OS, COST, ROST, RPE, and BM, are labeled. All images are plotted with identical color bar. an A-line rate of 25 kHz, as shown in Figure 6. Our rectangular The axial component of the scanner velocity induces Doppler modulation of the scanning beam reliably produced speckle- shifts to the interference fringe, thereby reducing the image reduced images in both mouse and human retinas without image SNR. To overcome this challenge, Szkulmowski et al. sug- blurring. gested minimizing camera exposure time (demonstrated at 5 μs) The ultimate target of vis-OCT retinal imaging is the clinical to reduce the effects of fringe washout. Such reduction of cam- adoption, where increased resolution and spectroscopic analysis era exposure time, however, would severely compromise image can improve the management of a variety of retinal diseases. quality in vis-OCT, which typically requires camera exposures Therefore, a primary design constraint of our speckle-reduction of ∼40 μs to achieve sufficiently high SNR in the human retina. technique was the usability by photographers in the clinical Our scanning technique does not introduce additional fringe environment, where advanced vis-OCT engineering skills are washout, as the scanning beam moves in discrete steps and unavailable. It is unreasonable to expect a clinical photographer is stationary during each camera exposure. Szkulmowski et al. to make optical adjustments to the system or troubleshoot tech- experienced ∼6-dB loss to peak signal-to-noise-ratio (PSNR) nical issues that arise. We implemented our speckle reduction in scan-optimized images at an exposure time of 20 μs, which in a clinical vis-OCT system at NYU Langone Medical Center they attribute to fringe washout. We did not observe PSNR drop simply via a software update. No additional optical calibrations in our scan-optimized images at an exposure time of 40 μs. were made to the clinical system, and no changes were made to Although our discrete scanning trajectory is optimized for vis- the photographers’ imaging protocol. Critically, the photogra- OCT, it is expected to provide superior image quality in other phers achieved comparable image quality (Fig. 8) to experts OCT systems with higher exposure times when using the same in a controlled lab environment and the clinical images showed sampling parameters as a resonance mirror-based technique. delineation of RPE and BM [Figs. 8(b) and 8(d)]. Micrometer- Our scanning technique also addresses the trajectory limits scale basal linear and basal laminar deposits between RPE and of a single resonant mirror shown in Szkulmowski et al. Using BM are thought to be early indicators of macular degeneration, discrete, software-controlled XY galvanometer scanners, we a leading cause of blindness. A future study in which these enabled multiple modulation waveforms across multiple coordi- morphological changes are observed in vivo in the clinic may nate systems (e.g., Cartesian and polar). There is promise for open a new window for the diagnosis and management of applying circular scan modulation for circumpapillary retinal macular degeneration. Such capabilities have been previously oximetry, where speckles in blood may disrupt true spectro- unavailable in clinical NIR OCT systems due to reduced axial scopic signal. The velocity of blood (e.g., ∼0.014 μm∕μs for resolution. Presently, NIR OCT is limited to imaging larger scale a 100-μm diameter human retinal artery ) is not high enough drusen, which are developed only at a more advanced stage of to completely uncorrelate the local scattering structure of eryth- macular degeneration. rocytes during consecutive A-lines (period of 40 μs). Therefore, Szkulmowski et al. performed averaging of offset A-lines A-line averaging across a regular circular scan is not perfectly in NIR OCT using a resonant scanner. One hardware challenge efficient for reducing the effects of speckle. By scanning a larger expressed in this study was fringe washout. In addition to the volume, our modulated circular scan is expected to acquire more complication of adding hardware to the system, the resonant uncorrelated speckle patterns in blood using the same number of scanner continuously moves during a single camera exposure. A-lines as a regular circular scan. Furthermore, the modulated Neurophotonics 041107-10 Oct–Dec 2019 Vol. 6(4) Rubinoff et al.: Speckle reduction in visible-light optical coherence. . . scan pattern is not expected to disrupt the SNR of the blood CNR and 2.87-fold increase in ENL can be achieved. We also signal since it does not induce additional fringe washout. showed that reducing speckle via direct B-scan averaging is not Next, by selecting a rectangular waveform, we best preserved as reliable for preserving the fine features, due to retinal motion. the lateral resolution along the B-scan axis and took full advan- Our speckle-reduction method offered a local sampling rate of tage of the space available along the orthogonal axis for speckle 3125 Hz (Sec. 2.1), which is well beyond the motion frequency. pattern decorrelation. In comparison, a resonant scanner can In the future, we will further improve the speckle-reduction only oscillate along a single axis. Furthermore, the reported res- performance, especially in clinics, by providing real-time feed- onant scanner uses a sinusoidal waveform, in which each mirror backs of CNR and ENL so that the photographer can adjust deflection along the orthogonal axis also carries a component imaging parameters to achieve optimal image quality for differ- along the B-scan axis. In this case, averaging consecutive ent eye conditions. In addition to exploring clinical benefits of A-lines will reduce the lateral resolution along the B-scan axis. discriminating minute anatomical features, such as the RPE and In addition, the sinusoidal trajectory makes orthogonal step size, BM, we will also investigate whether circular scan modulation d, nonlinear across each scan of the orthogonal axis. Our find- can improve the accuracy of measuring retinal oxygen saturation ings suggest that image quality is optimized when d is approx- because an srB-scan samples a larger retinal volume than a imately equal to the focused spot diameter along the orthogonal regular B-scan image. axis. Precise optimization is not possible with a nonlinear trajectory. Szkulmowski et al. partially overcame this by only Disclosures acquiring A-lines along the pseudolinear region of every other R.K., Y.W., J.S.S., and H.F.Z. have financial interests in sinusoidal edge, wasting a portion of light exposure on the Opticent Health, which did not support this work. Other authors retina. However, this is still less efficient than scanning with declare no conflicts of interest related to this article. a rectangular wave, in which all spaces along the orthogonal axis are linear and can be fully utilized. Acknowledgments Acquiring A-lines along the orthogonal axis increases imag- This work was supported in part by NIH Grants Nos. ing time by a factor of n ¼ 8. This limits the total number of R01EY026078, DP3DK108248, R01EY029121, R01EY028304, srB-scans per acquisition to 16 for an imaging time of ∼5s. R44EY026466, and T32EY25202. L.B. was supported by the Naturally, 16 raster srB-scans are not feasible to reconstruct a NSF Graduate Research Fellowship 1000260620. The authors high-quality en-face image of the human retina. A potential sol- would like to thank David Miller and Rozita Ghassabi for their ution is to decompose each srB-scan with n ¼ 8 into eight regu- helpful discussions. lar B-scans orthogonal axis, each of 1024 A-lines. If the spacing between each srB-scan is d, then we can decompose 16 srB- scans into 128 equidistant regular B-scans separated by the References distance d. By reducing the number of srA-lines per srB-scan 1. D. Huang et al., “Optical coherence tomography,” Science 254, to 4096 at n ¼ 8, we can decompose the full raster scan into 1178–1181 (1991). 512 A-lines −256 B-scans, and so on. The user may view regu- 2. M. R. Hee et al., “Optical coherence tomography of the human retina,” lar B-scans or srB-scans at a location of interest in the en-face Arch. Ophthalmol. 113, 325–332 (1995). image. Rather than choosing between a high-quality en-face 3. M. Wojtkowski, “High-speed optical coherence tomography: basics and image and high-quality B-scans, the user may adjust scanning applications,” Appl. Opt. 49, D30–D61 (2010). 4. J. M. Schmitt, S. H. Xiang, and K. M. Yung, “Speckle in optical coher- parameters to acquire both simultaneously. We recognize, how- ence tomography,” J. Biomed. Opt. 4(1), 95–105 (1999). ever, that the aforementioned solution is still not as optimal as a 5. M. R. Hee et al., “Optical coherence tomography of age-related macular regularly raster-scanned en-face image due to motion and that degeneration and choroidal neovascularization,” Ophthalmology 103, extra image registrations may be required. 1260–1270 (1996). 6. M. R. Hee et al., “Optical coherence tomography of central serous cho- rioretinopathy,” Am. J. Ophthalmol. 120,65–74 (1995). 5 Conclusion 7. A. Ozcan et al., “Speckle reduction in optical coherence tomography images using digital filtering,” J. Opt. Soc. Am. A 24, 1901–1910 We showed that our orthogonal modulation protocol signifi- (2007). cantly reduced the speckle for both raster and circular scans 8. M. Bashkansky and J. Reintjes, “Statistics and reduction of speckle in in vis-OCT without additional hardware. This protocol was optical coherence tomography,” Opt. Lett. 25, 545–547 (2000). robust against retinal motions. Furthermore, we showed that 9. A. Sakamoto, M. Hangai, and N. Yoshimura, “Spectral-domain optical speckle-reduced vis-OCT imaging does not require additional coherence tomography with multiple B-scan averaging for enhanced training for a clinical photographer to operate, allowing smooth imaging of retinal diseases,” Ophthalmology 115, 1071–1078 (2008). 10. M. Szkulmowski et al., “Efficient reduction of speckle noise in optical clinical translation. We designed this orthogonal modulation coherence tomography,” Opt. Express 20, 1337–1359 (2012). protocol to be compatible with arbitrary scanning trajectories 11. A. E. Desjardins et al., “Speckle reduction in OCT using massively- and demonstrated such capability in circular scans for the first parallel detection and frequency-domain ranging,” Opt. Express 14, time. 4736–4745 (2006). In the mouse retinas, we showed that up to a 2.35-dB (116%) 12. M. Pircher et al., “Speckle reduction in optical coherence tomography improvement in CNR and 3.1-fold increase in ENL can be by frequency compounding,” J. Biomed. Opt. 8(3), 565–569 (2003). 13. Y. Zhao et al., “Real-time speckle reduction in optical coherence tomog- achieved. We recognized that an anesthetized and stabilized raphy using the dual window method,” Biomed. Opt. Express 9, mouse retina had negligible motion, allowing for simple B-scan 616–622 (2018). averaging, and further conducted human retinal imaging. 14. O. Liba et al., “Speckle-modulating optical coherence tomography in Consistently improved image qualities after speckle reduction living mice and humans,” Nat. Commun. 8, 15845 (2017). were shown in human retinas using both an experimental and 15. C. Cuartas-Velez et al., “Volumetric non-local-means based speckle a clinical vis-OCT system and provide comparable results. reduction for optical coherence tomography,” Biomed. Opt. Express Here, we showed that up to a 2.25-dB (94%) improvement in 9, 3354–3372 (2018). Neurophotonics 041107-11 Oct–Dec 2019 Vol. 6(4) Rubinoff et al.: Speckle reduction in visible-light optical coherence. . . University, and a practicing ophthalmologist at Northwestern 16. X. Shu, L. Beckmann, and H. F. Zhang, “Visible-light optical coherence Memorial Hospital. She previously worked at the Jules Stein Eye tomography: a review,” J. Biomed. Opt. 22(12), 121707 (2017). Institute at UCLA and the Doheny Retina Institute at USC. Her 17. J. Yi et al., “Visible-light optical coherence tomography for retinal research interests include translational approaches to age-related oximetry,” Opt. Lett. 38, 1796–1798 (2013). macular degeneration and ischemic retinal diseases. 18. J. Yi et al., “Human retinal imaging using visible-light optical coherence tomography guided by scanning laser ophthalmoscopy,” Biomed. Opt. Xiaorong Liu is an assistant professor of biology and psychology Express 6, 3701–3713 (2015). at the University of Virginia. She received her PhD from the 19. S. P. Chong et al., “Ultrahigh resolution retinal imaging by visible light University of Virginia in 2002 and received postdoctoral training at OCT with longitudinal achromatization,” Biomed. Opt. Express 9, the Scripps Research Institute and the University of California, San 1477–1491 (2018). Francisco. Her research interests include the structural and functional 20. S. Martinez-Conde, S. L. Macknik, and D. H. Hubel, “The role of development of retinal ganglion cells and how they degenerate in fixational eye movements in visual perception,” Nat. Rev. Neurosci. glaucoma. 5, 229–240 (2004). 21. D. C. Adler, T. H. Ko, and J. G. Fujimoto, “Speckle reduction in optical Jenna Tauber is an MD candidate at the New York University (NYU) coherence tomography images by use of a spatially adaptive wavelet School of Medicine. She received her BS degree in communications filter,” Opt. Lett. 29, 2878–2880 (2004). from Cornell University. 22. B. T. Soetikno et al., “Visible-light optical coherence tomography oxi- metry based on circumpapillary scan and graph-search segmentation,” Katie Jones is a clinical photographer and lab supervisor at the Biomed. Opt. Express 9, 3640–3652 (2018). Advanced Ophthalmic Imaging Laboratory, NYU. She graduated in 2013 with a BS degree in biology from the University of Pittsburgh. 23. J. C. Booij et al., “The dynamic nature of Bruch’s membrane,” Prog. Retinal Eye Res. 29,1–18 (2010). Hiroshi Ishikawa is a professor in the Department of Ophthalmology, 24. C. A. Curcio et al., “The oil spill in ageing Bruch membrane,” Br. J. NYU and administrative director of the Ophthalmic Imaging Center. Ophthalmol. 95, 1638–1645 (2011). He received his MD from Mie University in Japan in 1989. 25. M. Guizar-Sicairos, S. T. Thurman, and J. R. Fienup, “Efficient subpixel image registration algorithms,” Opt. Lett. 33, 156–158 (2008). Joel S. Schuman is a professor and chairman of the Department of 26. S. H. Yun et al., “Motion artifacts in optical coherence tomography with Ophthalmology, NYU. He received his MD from Mt. Sinai School of frequency-domain ranging,” Opt. Express 12, 2977–2998 (2004). Medicine in 1984 and served as professor and chairman of the 27. C. E. Riva et al., “Blood velocity and volumetric flow-rate in human Department of Ophthalmology at the University of Pittsburgh from retinal-vessels,” Invest. Ophthalmol. Visual Sci. 26, 1124–1132 (1985). 2003 to 2016. He is a coinventor of OCT and a pioneer in the study and treatment of glaucoma. Ian Rubinoff is a second-year PhD student in biomedical engi- neering at the Functional Optical Imaging Lab (FOIL) at Northwestern Roman Kuranov is the head of product engineering and develop- University. In 2017, he graduated from Lehigh University with a ment at Opticent Health in Evanston, Illinois, and a research specialist BS degree in electrical engineering and physics. His research inter- at Northwestern University. He received his PhD in laser physics ests include innovating visible-light optical coherence tomography from the Institute of Applied Physics of the Russian Academy of (vis-OCT) technology, optics, and signal processing. Sciences and did his postdoctoral training at the University of Texas, Galveston. He previously worked as principal scientist in Wasatch Lisa Beckmann is a third-year PhD student in biomedical engineer- Photonics and as an instructor and researcher at the University of ing at the FOIL at Northwestern University. In 2016, she graduated Texas, San Antonio. from the California Institute of Technology in biomedical engineering. Her research interests include using vis-OCT to study eye patholo- Hao F. Zhang is a professor in the Department of Biomedical gies, optics, and animal model imaging. Engineering at Northwestern University. He leads the FOIL, which seeks to develop and apply new optical imaging technologies for Yuanbo Wang is a senior project engineer at Opticent Health in biological study and clinical use. His lab pioneered the development Evanston, Illinois. He graduated with a PhD in biomedical engineering of vis-OCT for high-resolution and functional retinal imaging. He from the University of Missouri-Columbia, where he studied polariza- received his PhD in biomedical engineering from the Texas A&M tion sensitive OCT and fiber tractography. University in 2006 and received his postdoctoral training from the Department of Biomedical Engineering at Washington University in Amani A. Fawzi is the Cyrus Tang and Lee Jampol Professor of St. Louis. Zhang also cofounded Opticent Health, which seeks to Ophthalmology at the Feinberg School of Medicine, Northwestern translate vis-OCT to the clinic. Neurophotonics 041107-12 Oct–Dec 2019 Vol. 6(4)

Journal

NeurophotonicsSPIE

Published: Oct 1, 2019

Keywords: optical coherence tomography; visible light; speckle; retina; imaging; clinic

There are no references for this article.